PyLadies Fireside Chat

Speaker 1 [00:00]

blamers for the session. Please make sure your mobile phones are on silent so that there are no distractions during the session. And, yeah, whenever you have any questions, you think of any questions while the session is going on, then just write them down in the talk.de, PyCon talk.de, as usual. Yeah. Yeah, we're ready? Okay. So the session here is a plenary talk, and it will be about PyLadies, a fireside chat, in case you were wondering where you are at the moment. Let me introduce our panel. So, we have Teresa Yoshu, a leadership coach, and then we have Dawn Wages, exactly, from Anaconda.

Speaker 2 [00:53]

Jessica Green.

Speaker 1 [00:53]

Jessica Green from Ecosia.

Speaker 2 [00:54]

Jessica Green.

Speaker 1 [00:55]

So please put your hands together to welcome them on stage and I wish you a very nice panel together.

Speaker 3 [01:06]

Thank you, so welcome who here was last year at our panel

Speaker 4 [01:15]

Ah, someone's turning.

Speaker 3 [01:16]

Okay. So today we have two special guests. From the Python community that you may or may not know. So we start already with the introductions directly. So maybe if you want to tell us, maybe Don, you first tell us something about yourself. What do you do with Python today? Awesome.

Speaker 4 [01:38]

I started out as a Python generalist with Django and now I'm exploring more data science and AI ways and I'm really excited to hear from Jessica and team and maybe some questions will come up on how people are using ethical and feminist AI practices so I can learn, grow and build with you all. I do developer relations so I teach as I grow.

Speaker 3 [02:06]

Yeah.

Speaker 2 [02:07]

Yeah, I'm Jessica. I use she, her pronouns. I'm a machine learning engineer at Ecosia, which is the search platform for a better planet. I am also one of the co-organizers for PyLadies Berlin chapter. I sit on the Python software for band board, and I am also the co-chair for the Python software foundation conduct work group.

Speaker 3 [02:32]

Cool. Thank you. So maybe like a side question, just because I guess a lot of this information we can find about you on LinkedIn. What did you do before Python?

Speaker 4 [02:42]

I also want to hear about your background and maybe the audience want to know. Can you introduce yourself too, Teresa? Okay. Sorry to take over.

Speaker 3 [02:48]

over. Bye. No, okay. So I'm Teresa. My pronouns are she, her. And I live in Hamburg. I am

Speaker 4 [02:49]

Bye.

Speaker 3 [02:55]

currently a data leadership coach. And I've been working in data something, something for the last years. And I'm also doing PyLadies Hamburg. And I was involved in a lot of the PSF group work, diversity and inclusion work group, code of conduct.

Speaker 1 [03:12]

conduct

Speaker 3 [03:13]

So if you've been naughty, I probably know about it. And, yeah.

Speaker 2 [03:20]

Also on the board.

Speaker 3 [03:21]

the board. I was also on the Python software Ferbant board, and this year I think I took a step back and I left all of the spaces just to take a moment to take care of myself, and I will come back at some point. Yeah. Cool. Okay. So, life before Python, what was it like?

Speaker 4 [03:43]

As I was learning Python, I worked at a cheese steak restaurant, I was learning how to be a data analyst, I have an undergraduate degree in business, and tried my hand at finance and it did not work. I did not like finance and finance did not like me. And so I was really one of those people where I was kind of learning in the open with friends because of my Python users group in Philly, and started upscaling there. And you all mentioned your backgrounds with the PSF, and I kind of skipped over that a little bit, but fast forward about a decade, I was the chair of the Python Software Foundation for about three years, or two years, and then the treasurer. And that was my first foray into really doing leadership for open source. I was a co-founder of Django Not Space, the mentorship program, it is running so smoothly, and I'm not part of it anymore, but I'm cheering them on, it's so fantastic. So if you are interested in ramping up as a leader in the Django community specifically, I'll just do a little plug for them. But yeah, it took me there, that was my life before Python.

Speaker 3 [04:58]

Awesome. We have the same plug for the PyLadies community, of course. Same side effects, you learn leadership and all sorts of other things besides Python.

Speaker 2 [05:07]

and

Speaker 3 [05:08]

And Jessica.

Speaker 2 [05:09]

Jessica? Yeah, I mean, I've heard only great things about the Django Nort programme, so clearly laid on the foundations of giants, I very much admire your work. So I have a little bit of a diverse background, I guess, I'm a career changer, I studied film production and worked in film and television. If anyone knows the TV show Sherlock, I worked on the pilot of this and various other stuff. Then when I moved to Germany, I ended up working in specialty coffee for five years as a coffee roaster, and because I guess I'm an indecisive person, I then decided to switch and I changed into tech, and I consider myself very much a self-slash-community-taught developer. I came across very early on the PyLadies community in Berlin, and even though at that point Python was not a primary language for me, my first job I mostly programmed with Go. It was really the community that I kept going for. I have then kind of transitioned more into machine learning where I have ended up using Python more and it's maybe always been like a good second language. Now it's a lot of TypeScript, I have to say. But I'm very grateful for the community, specifically the Python community, for always being there as a network and to support someone like myself. I often feel very intimidated at not having this more academic or traditional background, but I think I've managed to really find a space for myself.

Speaker 3 [06:45]

Cool, thank you. So we have the question that is on the website about this event. So I think we have to ask it. And that goes like, what does it mean to build with Python when AI is reshaping everything? Any thoughts on...?

Speaker 2 [07:05]

Would you like to go first? I mean, maybe I can just dive into what I just said around that I predominantly now am using TypeScript. So I think it's very interesting what a bump Python got as a language, as a programming language, and as a community when everything kind of switched towards machine learning or more specifically large language models. I think it gave a lot of exposure to the community, and this was mainly because of the diverse ecosystem that it had already built for data science, but also that it's like very fully open source and kind of the community is there in a very supportive manner. Unfortunately, I have to say what I'm not seeing is a lot of the benefits of that going back into the Python community, which is somewhat problematic. It adds to this image of expectation through AI and a lot of the things that Karen Howe talks about in her book, Empire of AI, I think we can make that analogy here as well. I think one of the things when I was learning to program, because everyone has opinions on programming languages, I'm sure you do too, and I remember Python was always kind of like, yeah, well, it's a good second language, it can kind of fill the gaps, but it doesn't really excel in this area or that area, maybe it's not low-level enough for really performant backends, and if you're building on web, at some point, no matter how much you dislike it, you will have to touch JavaScript. I know that everyone who tries to use Wasm and so forth is desperate to get away from it, but it is really the lingua franca for the web. And I think that's now where we find ourselves in terms of as things move more towards actually calling APIs, rather than hosting or building models, you don't actually need all of the packages and the ecosystem of Python as much, and it's very possible to do this then in different languages. So I think it's a very interesting point for Python. I think we gained a lot of exposure and a lot of benefits also from that. But I wonder where this will change in the next five years.

Speaker 4 [09:07]

It's an interesting approach, and I see that a lot of when there are first APIs available, the first they're supporting is Node and Python. But I do love that Python remains first, and in my world, I am staying very, very much in Python, and if I am expanding to other languages, it's going into Rust or C. So I I appreciate the evolution of the community in that space. And I stay mostly in Python, so one of the things that I've noticed, and I think we'll also get to it, is the way that the career has changed in the ecosystem and the expectations of that. A lot of times, even the most AI reticent of programmers are using AI assistance while programming or using code generation tools and it expedites a lot of that feedback loop that you would otherwise go to documentation or to blogs to get the best opinion on architecture paradigms so one of the things that I will do while also being thoughtful about compute and Jessica I would love to hear your thoughts on compute resource and what is a worthy task and we were even joking about kind of asking Claude and did we use the answer later and was it worthwhile computer for forgetting to look back at our chat kind of thing and I really want to be thoughtful but machine learning has existed for decades I mean it is built off of linear regression and mathematical computations and HPC innovations that we that we did before so now that we're able to do a lot of that locally and so that's another point of it is there the feedback loop of programming but also I have gotten a lot of efficiencies with focusing on local first development which has been really exciting and it makes me feel more in control of my stack and of the compute resources that I'm putting on the environment but the chip has to be made there's silicon silicon other tools that have things components that need to be mined And there's sometimes a backlog and also financial considerations, right? Like not everyone can have a worthwhile enough machine to be able to run some of those local things, right? I loved starting, being able to instruct someone to start with a Chromebook and get started with their web dev journey in Python. So I think there's a lot of changes on the hardware and infrastructure level that's going to be able to democratize the access to AI. And that's not just universal compute credits, okay? I am not shelling universal compute credits, a la some of the famous founders. So a lot's changed, and I think we're all trying to keep up.

Speaker 2 [11:59]

Yeah, no, I definitely agree. And I think in terms of this idea of democratizing things and like being able to make things more accessible, there is like a lot within that in terms of like the open source models and the ways that you can like run these things on smaller and smaller machines. and it's interesting if you haven't been up to the feminist AI LAN room they have all of these things running and they will also help you run them and you don't always per se need to have like the beefiest laptop and all the compute no matter what some folks would tell you otherwise but then it's also like how to compete and how to deal with this current uncertainty in the workplace and in the market and how can people really kind of build something for themselves that feels more future-proof. It's a difficult landscape right now.

Speaker 3 [12:55]

Cool. So I wanted to talk about the change word that you used and a little bit thinking about that you all were a junior at some point and maybe several times in your career, in different careers. And I think this is a little bit like, well, I like to say working in data, at some point you know how to predict things. But the question is a little bit, what do you see that, what will a junior developer look like in three years from now. Interesting.

Speaker 4 [13:27]

I can start. I work with a lot of junior developers and developer relations. Just kind of a plug on what developer relations means, because it's not always an accessible kind of job title for people, is I like to consider it a feedback loop with the ecosystem. So, yes, I'm paid by a company, but when they have the forethought to set up a team member to just go out into the community and teach people through their tools, sometimes you're able to unlock a lot of core competencies learning through a tool, but you don't necessarily need to remain reliant on that tool. So a lot of Anaconda's infrastructure, I do developer relations for Anaconda, is lower level. You're talking about maintaining Python environments, which is the first step for a lot of people. And then also having architecture or, you know, compute considerations. What GPU are you running on at all or a CPU or what machine or what operating system? So Windows, Linux, and Mac. And so those are the first steps, and people don't necessarily realize that they're encountering really technically complex things on your first step. When you are PIP installing, condo installing, UV installing, you are building off of the shoulders of giants, as we were talking about before. There's a lot that came before. And when something breaks, it is really easy to internalize that. And one of the things I really like about chat agents, especially when we give them names a bit, and I don't like typically the personification of these inanimate objects, but it does alleviate some of the shame or guilt to have a pair programmer right next to you. If something breaks in you without having to internalize that, copy the error and put it somewhere else, and it's not just Stack Overflow shaming you because there was a whole kind of error of Stack Overflow kind of being a bit toxic, but you have this very friendly named agent next to you telling you, no, it's totally okay, and here are some resources on how to fix it, and then it actually works, that can absolutely supercharge a junior developer. so as long as I think we're again we're evolving the industry a lot and so I want to do that and consider it and thoughtful ways to the environment and to my peers and to the tools and to the ecosystem and to the language but I think the junior developer is really going to focus on system level and architectural things so buy those books on system level design and good architecture and data-driven design and you're also going to be focusing on the domain space that you are applying code in so understand if you're building software for finance learn a little bit more about the user experience of the software versus the technical implications on how to understand the syntax and that's still very technical you're going to want your object-oriented programming, you're going to want your models and your architecture to be proficient in that environment. So extending that layer a little bit more. Sounds meaty. It was kind of the domain of the senior engineer, but that can go a little bit lower into the career stack, I think.

Speaker 2 [16:58]

I agree. I agree. And I think if you're a senior or a principal engineer, do not gatekeep these skills. And we have to be really mindful of the world that we're playing. It's not a nice market for anyone. So I think it's very understandable if people want to make sure they're still being valued and recognized and keeping their job. I think we all would like that. But the skill shift is going to move and we need to make sure that we're not blocking it if possible.

Speaker 3 [17:25]

I wanted to have a follow-up question on the Claude and Chatty and all of this. I call it Chatty. Anyways, they are very nice. And I think it's raising the expectation on how other people should behave at work as well. So a lot of the time, you know, sometimes people spend time in a specific culture at work and we tend, people get older or more experienced and then sometimes we have toxic colleagues and so on. What is the future of the toxic senior engineer when Claude is knowing everything that they know but is polite?

Speaker 4 [18:02]

I'm happy to take a spicy question first, but Jessica, do you want to take a stab at it?

Speaker 2 [18:05]

take a stab at it? I mean, if you... No, you go, you go.

Speaker 4 [18:08]

They're going to go the way of the dodo birds. I mean, there are, so I'll take a step back and think about, like, team and management culture in general. There are studies that say that a toxic personality at work will ruin the environment for many of your engineers and absolutely decreases, in measurable ways, productivity. so do I know do I know that the sycophantic but in a better way more like um praising or kind language of our chatties are going to upskill everyone I didn't think about that yet but that's a really good question I mean the studies are

Speaker 3 [18:48]

I mean, the studies about toxic behavior at work have existed for a while now. Yep. Just nobody kind of cares about it.

Speaker 4 [18:56]

And it, or puts.

Speaker 3 [18:58]

money where the...

Speaker 4 [18:59]

yeah and now we're connecting the dots on things I really like that idea I hadn't thought about it

Speaker 3 [19:00]

Yeah.

Speaker 4 [19:04]

but yeah I think getting a toxic personality out has always been my objective and whenever I'm in a workspace of like this is either they leave or I do and so I have a very low tolerance for working with unproductive and negative or unkind people and you know directness is always something I love and there's a fine line well actually I think there's a pretty thick line between being direct and being a negative person so I love that that question I hadn't thought about it before and now I'm going to go noodle on it after this

Speaker 3 [19:39]

You're welcome

Speaker 2 [19:42]

Did you have thoughts, Jessica? I mean, I just hope it's that outcome over reducing more human connection. I think, like, the thing that you see promoted a lot is the idea that AI is making it a more even playing field and more open to more people. And then if you look who's actually getting money, it's not so different. So I like the aspiration that this can have a positive impact. I'm seeing little evidence.

Speaker 3 [20:17]

evidence maybe it's something that we can use it as an answer whenever someone is rude it's like you know what you're absolutely right i will go ask claude he's polite

Speaker 2 [20:26]

Ha, ha, ha, ha.

Speaker 3 [20:29]

Be more open about comparing someone to a chatbot if that's what they want to be reduced to. Sorry. Digressing. Digressing. Yeah. Okay. So, let's, going into a more, another polarizing topic, if, like, looking at your current job, the work that you do as it is, let's say, in how many years do you feel AI could do it with a little bit of improvement, a little bit of instructions from yourself? How long do you feel like you could, how much work can you invest in delegating it? Yes.

Speaker 4 [21:14]

So there's two ways to DevRel. Actually, there's a million ways to do developer relations. But one of the lenses that I navigate with my colleagues is being a content engine, right? Like you go to DevRel to just be more of the social media or content asset. But if you are building scripts or writing blog posts or even creating videos, right, with the right models, you can mimic a face and voice and or just have a voice over top of screen shares and have I used it I've just personally decided to not use image generation or video generation through AI tools because that's my personal line on what's worthwhile for compute so I don't tech I don't do that although I love AI for translation services, but then multilingual dev rel now may experience a bit of competition in that particular bullet of their resume. So when I dev rel, I emphasize the relationship building. And so you can't have extemporaneous conversations with a chat that's really going to dig down and draw connections sure there's pattern recognition but nothing's going to beat the pattern recognition of like authentic human focused conversation and so where my manager will benefit from having me and additional engineers when I talk about like hey I want to grow my team this is why and not just throw a chat at it or some image generation on it it is building those connections and having us all go out and bring authentic stories to it you may want to listen to me because i said i used to work at a cheesesteak place and i was learning python in my nights and weekends um and not because i am this soulless chat that is a bit sycophantic and will search things for you and just bring you the answer in a second we can find the answer together and we can build community together, and I think that's way more valuable if you want to make someone a super fan of your tool, if they really want to be focused on the company outcome. I'm trying to build super fans, and I don't know if a chatbot is going to be able to create a super fan.

Speaker 2 [23:35]

I think it's a bit different in my field, and I think it's more of a shift than a replacement. Like we talked about before, I think a lot of the skill sets that maybe were further up the line are coming further down, and also a lot of the, like, business understanding, product understanding, domain expertise is also becoming more relevant. I'm definitely not in the opinion of developers are going away in like two years or such. I think we see now even like more demand for software. We definitely are having like an interesting moment in the market in terms of input of like new talent. So it's gonna be curious in a few years when we have no juniors that want to be mids and no mids that are getting into senior roles. But I think this also has to change, to be honest, at some point. I think what I wanted to comment on was like what we talked about in terms of like junior roles and like how things are changing. I think also like knowing more about platform and where your code's running is something that started to happen when I also got into programming. So I remember a lot of the time when I was doing backend engineering, I was also doing a lot of the infrastructure and like DevOps, essentially. And I remember talking to a lot of people who did more pure back end in terms of, like, we write this and we pass it on to someone else. And I think this idea of shifting responsibilities, becoming more of a generalist or adding additional skills has kind of been going on for a while. It's not per se new. Maybe it's accelerated with AI. And I think the other great thing around juniors or people coming in is that they're a bit like blank sheets. So I remember when I joined the tech industry, I would hear a lot of people talking about, like I said, languages of choice or this, that, and the other. And to me, I was just like, I don't know about this. I don't really have an opinion on it. So I'm really happy to just build and explore and be curious. I think this is still the most important skill, to be honest, if you're getting into the industry is like that curiosity, that desire to build. A lot of these tools can actually support that rather than hinder it, to be honest.

Speaker 3 [26:05]

Cool. Okay. So maybe one more, I think... So we, by the way, for everybody, I would like to invite you again to submit questions. If you haven't, we will have for this fireside chat quite a lot of space for questions, so it's not just the last five minutes. So do have as many questions as you want. And from the learning Python, and I think still because we are at a Python conference, and I heard some people saying what is the future of Python conferences, do we still need Python conferences or we just call them coding community conferences and so on? Maybe a little bit going into how would you, if it were you approaching, with everything that you know about everything so far, how would you approach learning Python today?

Speaker 2 [26:56]

Yeah, I can go. I mean, I think building things. I really love the book Automate the Boring Stuff with Python by Alice Swakart. That's always my go-to recommendation for anyone because it's really about being practical hands-on, building something. I think also what I really enjoyed also around MicroPython and learning how it plugs into C is optimizing. And something we touched on before around like this whole like cloud environment and that everything is kind of like not tangible in a way because it runs in the cloud and like cloud is, cloud compute generally speaking is fairly cheap and also just exists somewhere. But now we are starting to see a lot of more like on-prem stuff, which is super interesting on device things like embedded. And I think this is a very, like, personally, what's exciting me right now is, like, robotics and trying to think about this. And I think Python is maybe not the most low-level language in terms of being able to, like, really touch the board and, like, what's going on. But you can still get a feel for this and use Python for these kind of things because it's so versatile. I mean, it's been in space, so.

Speaker 4 [28:17]

100% I would just yes and all of that I'm so I mentioned that I start with local first development as my kind of thing and I'm because I'm in DevRel and I explain myself a lot then I am now adopting it as a brand and every time I check in with someone to have a conversation I'm like yeah I believe this more yes I am a local first AI and ML developer or engineer of sorts And then I've been really excited about Wasm. You've mentioned Wasm as well, too. And so one of my team members, just to kind of give a shout-out to Beware and PyScript, two of my team members are contributing to CPython to make Wasm a first class eventually, but a platform for Python. And so that unlocks so many different capabilities. So Python could go on your phone. I mean, it already is, but even more performantly, and how do you get it on a chip or on a small portable device or putting Python on a vape? I've seen someone experiment with hacking into a vape after it was done and using the chip in there. Also, something I'm really excited about with physical hardware, and it's interesting how you are contrasting this cloud, which is nondescript over there. I know what region it's in, right? Like, I can put a dropdown menu and know where roughly my cloud might have an outage. But now with that shift more in on-prem or on-device, I'm also kind of being more considered about batteries as well. And so this is really getting into the engineer. But I have a friend who was a chemistry major when we went to college together and now is like a battery expert. And I believe that figuring out that battery infrastructure, and that is not me, that is me admiring an entirely different discipline, I think it's going to unlock really amazing things with the portability of compute and the availability of data centers and the damage that data centers might do. And so I'm just watching that space. The hardware, I feel like, is going to be really fun in the next several decades. I don't know how fast it's going to make for them to do it, but I'm waiting with bated breath on what they come up with. I don't even know if I answered the question. I just went on a rant on something interesting.

Speaker 2 [30:48]

It's a fireside chat.

Speaker 3 [30:51]

I have a last question about Python and careers and so on, and maybe it's a little bit on what is a decision in your Pythonic career that you've been putting off for a while now?

Speaker 2 [31:04]

What is a decision in your Pythonic career?

Speaker 3 [31:08]

Yes, I have to I scoped this yeah, right

Speaker 4 [31:10]

Yeah, right. I did, we chatted about it, so I do have one at the ready, so I can... Oh, please. So this is a bit vulnerable, but, like, really, I learn in the open. It's part of the way that the most accessible people I... I'm doing a preamble, but the most accessible and expert people that I love and learn from in the career are really vulnerable on the things that they are experts on and the things that they're not, and so I mentioned that I have kind of like a fangirl over WebAssembly and some of the long-time career investments some of my peers have done that I work with at this job and my last job. And it's been a long-term goal for me to contribute to CPython, some of the WebAssembly support that I see my team members do. And there is a long list of issues that some people are working on. Brett Cannon, among many of them, Nicholas Tollervey, and Russell Keith-McGee are just lovely legends in the community. And Brett Cannon was the person who authored or who mentioned come for the language, stay for the community. He's a legend. And I've just not been able to do that in the past couple years. And it's mostly because as a marginalized woman, as a woman of color, queer woman of color, I am really being gainfully employed and well compensated is just a thing that keeps me up at night. And having a career gap or focusing off of the thing that makes me money to focus on open source and something deeply technical has never been something I've even had the capacity to indulge in before and now that it is I'm luckily enough in a place in my career where I can take my eyes off the ball a little bit and be a little bit safer with my career and focus on open source it is almost like I'm too scared to do it I am at my terminal I have an open issue and then I get a ping on slack from work and I'm like let's focus on the thing that pays me. Um, so that's one of the things that I've been really hesitant to get into, um, and like focus on. And it's not necessarily about whether or not I can do it. Unfortunately, I have a very large ego. I know I can do it. Um, but it's, uh, and we all have our different things. I'm a cheerleader for other people getting big egos too. Um, uh, but it's, it's really like making the time in figuring out the personal validation and motivation for doing it. So yeah, that's my thing. And maybe you'll find out after this. I mentioned it on the stage and then I get my first PR merch. I don't know. But like watch this space. Maybe we can cheer along with me. It just hasn't happened yet.

Speaker 3 [33:46]

I mean, Claude is going to watch this talk, and he's going to give you next time to ask it something. He's like, shouldn't you do this other thing? Yeah. Okay.

Speaker 2 [33:55]

I mean, co-sign that statement. I also feel very conflicted in terms of how much capacity I have for things outside. And as I mentioned now, I'm working a lot with TypeScript, which is a language I also don't super know well. And should I be shifting my focus on improving this? Or should I be leaning into the things that bring me joy? This is not a comment on TypeScript per se. is just, like, the things that I would prioritise for myself. I think it's a very fine time to just hold space, quite frankly, and it's something I try to model and try to, like, communicate in the community as well, is it is really difficult to be ambitious right now because it's a time of great uncertainty and a lot of people are very unsure how things are going to look and they want to be able to pay their bills and maybe have some money afterwards and that's totally fine and I really identify with what you say of feeling like paralysis on terms of like well maybe I'm at the point where I can do this but I feel blocked because I'm unsure if it really is the right time so I definitely feel that and I think my message to people would be like it's very okay and like I think one of the things that we talked about or that you mentioned on was a little bit around this thing of like intentionality when should we use llms when is it when is it wasteful maybe so to speak because we're not actually using the output um and i think this we have to think about this in the realm of like who we are and where we are in the bigger picture to be quite honest so like if you are starting out these tools can really help you if you don't use every response honestly you're not doing as much damage to the world as the people who run the companies that own these LLMs and I really think you have to see that and think about it in a measured way it's like a lot when we talk about climate change and we talk about the impact that individuals have versus the impact that huge corporations have it's not comparable and I'm not saying like okay just like you know throw it to the wind and do whatever of course that should look how it needs to look for yourself for you to like align with your values of what you want to actually be putting out into the world but I do think we put way much to onus on the individuals and I think you know there's alternatives to play out with out there like running models locally you're seeing this is becoming easier and easier it has like anyway other benefits of like data privacy but also being able to use it when you don't have the internet for example but I really don't feel like people should feel shame in these things most people are using the coding assistance most people are just calling, opening an eye. Every conference I go to, I look to see, and this is facts. So, like, we don't need to hold ourselves back, especially as, like, people from marginalised or underrepresented groups. We need to make sure that we're participating in a way that is intentional for ourselves and matches what we want to do. I'm sorry. Oh, good.

Speaker 3 [37:13]

Fireside chat. So now we are having questions from the audience

Speaker 1 [37:18]

Thank you so much for such an insightful panel. Please, an applause before we start the talk. That was really valuable, really. So we have a handful of questions, and I will take the one that is upvoted. So if you want one of your questions to be up in the list, please upvote them now. Yeah, so the first question is, it feels like we're headed for a new feudal system where companies, communities, and individuals may thrive as long as they pledge allegiance. For example, paying big money to one of the AI lords. Do you think this is an overly pessimistic outlook or do you think local first AI may keep up?

Speaker 4 [38:00]

I'll say, because I call myself a local first AI engineer, so I'll start with that. And it paints a really interesting picture, right? Like it's visual in my mind, and I'm thinking about, like, these sects and kings and... I think that has been there for a long time already. And AI, like all things, is a force multiplier for all systems that currently already exist. But eventually, those systems bend until they break. And I am incredibly inspired by things like the Feminist AI Land Party and the work that I see from different countries and regulatory bodies and standards bodies that are setting the goalpost on what good looks like. And I will eternally be an optimist. And I think that the temerity of the human spirit will prevail and corporations will continue to have to pivot towards us. And it is a farce that they have control of it. They are shaking in their boots every time we claw power back. and there's power in the numbers so i i believe that the the decision matrix is different um but i maybe and i don't know if i have a ton of evidence to support this i mean the marketing dollars are saying something different but uh i i believe that the we will break free

Speaker 2 [39:51]

Yeah, I think things swing back and forth, and it's, like, honestly beyond so much more than technology or the tech industry, to be honest, that, like, leads this question. I mean, I think the big reckoning right now is, like, the main thing is the cloud companies, and this is the ecosystem on which these things are running. so I would agree that it's less AI and more an existing platform that of course is being hyped through the AI lens. Something I'm thinking about a lot at the moment is complicity and you know I think the thing is like we're all

Speaker 4 [40:33]

all

Speaker 2 [40:34]

as tech workers to a certain extent complicit with this setup so it's easy for us to be like it's a them thing and they're leading this but pretty much also if you think about the Me Too movement it's also about the system that props up these things and allows for them to be the case and I'm thinking more and more about how we can intentionally show up in ways that question this complicity but also allow us to live with the conflict of the fact that we work in an industry that is propping this up and it has so much damage beyond just ourselves and the workforce within it.

Speaker 3 [41:17]

I kind of had in my mind an idea of how streaming, like movie streaming industry went. As an example, we have all the big players, but I'm pretty sure there are still people. I still know people that download their own movies from somewhere and they're not streaming from the main providers. And in the end, it's a matter of convenience. If you're feeling lazy about figuring out how to buy things and how to do it yourself, you are going to just pay for that convenience and you're giving away other things with not just the money and in the end I think a lot of people there are still some that are running things locally but

Speaker 2 [41:59]

But it's not just laziness. It's also like there's a huge amount of lobbying and money put into ensuring that you buy into these ecosystems as consumers. I mean, it's capitalism. Let's call it what it is.

Speaker 1 [42:12]

And speaking about convenience, there was a question about, given your experience, what would you say to people who are afraid to take the first step in AI?

Speaker 2 [42:24]

I mean, it depends why they're afraid. So, yeah.

Speaker 4 [42:29]

What would you say, given the caveats? I mean, maybe the typical responses you think someone may be scared.

Speaker 1 [42:36]

Any suggestions?

Speaker 2 [42:36]

I think the thing is we have to work with and accept that some people just don't want to use AI. They don't want to use such a large system that causes damage. And there are then maybe smaller alternatives and helping them set up local prem and so forth. But there are people that just are like, I don't want to do this. And I think that's completely valid. but then we also need to make sure we work towards spaces that incorporate that as well and keep humans in a space where they can show up authentically for work if it's more to do with I'm not sure how to get started or I'm intimidated by this I do think there's some great resources and spaces like PyLadies we have a lot of workshops in Berlin but also of course in all the other chapters around the world there are plenty of things now online around this and just finding people who you connect with who are doing these things and working in the open and like you were saying Dawn I think there's more of us that should be doing this to be honest that's not calling you out it's calling me out too of like just like really breaking down the barrier of saying these things are systems just like any other system and not making them into like this endless black box of magic.

Speaker 3 [44:04]

They're just tools just tools just tools you use them To it. I think there's a little bit of an elephant in the room There's a lot of people that will say that it will just make you a lot faster But I think there's a lot of work into making AI making you faster now Most of the time if you just use it it actually slows you down

Speaker 4 [44:22]

That's actually a really good point. There is a, gosh, I wish I could cite the study, but go fact-check me, please, and tell me if I'm wrong. But I'm pretty sure there's a study that said that it doesn't, so there's an idea that is propelled by some of the owners of the large AI systems that it will remove the amount of work, and thus we will have more freedom, and they act like that is the thing that will unleash us from capitalism and we're finding that is not true that just expectations get larger and I think we've been hinting at that here of we're you're not finding that people are actually reducing their amount of time that they're standing sitting at the keyboard they're just doing more and have more hats on and are expected to act more dynamically and in a good way and like there is a component of that that's a really good but also there we need to like you mentioned Jessica make space and be intentional and understand our tools and it is not a black box it is just a tool accelerator and you know drink water go take walks like sometimes it's about getting back to the fundamentals and in response to the question on getting started and I absolutely co-sign and I don't know if we even have much more to add, that it's just a tool, it is not magic, and trying to approach it as a tool may make it more digestible or feel more feasible. What is the vector at which I'm using it? Okay, I'm using it in a browser. What does it take to get my AI in this browser? Interesting. Okay, so where is the data coming from? Data ingress, data egress. Okay, so there is always going to be some of these fundamental software principles or just systems principle complex systems principles that are going to be applied to your AI as well and breaking it down into parts maybe making yourself flash cards bringing friends along for the journey may make it a little bit more digestible

Speaker 2 [46:26]

Yeah, and I think it won't hit every domain in the same way right like the thing is I work in web web dev and I think this is an area where it's more much more sufficient and optimized for based on the training data and the type of work that that actually entails but there are niches where this is probably less.

Speaker 3 [46:47]

Yeah, for example the water damage at my house is still that they don't find workers Yeah, we still have jobs that are yep not gonna be done with AI

Speaker 4 [46:58]

I love the meme that I see online, and it's like, could AI do this? And it's just like random things, people doing a weird dance, people making an odd joke. And it's like, no, AI actually couldn't do that. You absolutely have value. And it is the way memification of real things kind of shows up. But yeah, absolutely.

Speaker 2 [47:16]

Yeah, absolutely. I mean, it's also what we're talking about is what is AI, right? I think at the moment that topic is really around like some sort of large language model that is built within a system that has access to other information and probably tools if we want to say it's agentic and that it runs in some sort of loop or if statements. And I think this is the thing of demystifying of like that isn't actually the whole topic of AI. That is a variant that is very popular right now and is receiving most of the attention and most of the research grants. And probably what is coming is putting this into hardware and robotics because you can see that all of the companies that lead this space are building hardware because they want to also control where these things are run so that they can also have more insights into our data. I mean, I very much think it's surveillance technology and that's why it's important to explore routes where we're not continuing to accelerate giving our data for this expansion of capitalism.

Speaker 4 [48:26]

Both individuals and companies are really interested in data sovereignty, and I find a large hub of that in Europe as well. Like, there is much more conversations, but it's all over the world. But whenever I come over here, there is this discussion of keeping data and AI sovereignty on your AI stack. And I love how you kind of broke down really what is AI, because for a while, I mean, for the first few months or years, everything's moving really fast. It was like, there is no such thing as AI. It's hype cycle, it's machine learning, but now we have this language around it and it's too sticky to remove, I believe, and I think it's actually, in a certain way, when we all understand the same word in the same way, it's helpful to continue to call it AI, but still being thoughtful and poking holes at it. And the last point, because I know I keep talking about this and I don't want to run too long, but there are a few points, vectors, which individual incentives and corporate incentives align, and those are the few ways that you can kind of jump over and leverage their tools. If you are interested in AI, the hyperscalers, the Microsofts, the NVIDIAs, the AWSs, the Googles, do want you to learn AI, and so they do have lots of resources to build AI and ML engineers, and many of them have free courses and certifications, not all the certifications are free, so you could learn there. And so that is one vector at which we're all kind of going in the same direction, if that's what you decide. But then, really, we've been discussing a lot, being incredulous of each of the steps and understanding where, and they're probably going to want you to put it on some type of remote compute, and so then being thoughtful on, okay, what's the next step? They're asking me, they're telling me to do it this way, that's great, that's a great primer, let's challenge this and ask more questions around why they are giving me this information.

Speaker 3 [50:26]

I just wanted to add to the naming that the sticky name, like AI, doesn't actually have to stay in our mind as artificial intelligence. I heard a very good alternative name is almost intelligent.

Speaker 4 [50:39]

I like that one.

Speaker 1 [50:42]

That's a good one. Okay, we have time for one last question. So, there seems to be a change in how people learn in the age of LLMs. Juniors will often prefer chat to Docs or Google, let Cloud Code build something, and then understand what's going on, et cetera. Do we need or have some specific best practices or guidance to avoid possible anti-patterns, like learning, internalizing something wrong from slob code or chat debugging? Do you have any guidance for this?

Speaker 4 [51:15]

This is a lot of the workflow because that is education in an unconventional way, right? Like I'm not a teacher in front of a classroom all the time, although I will get invited in to do like a single lecture. So even as the producer of content, right, like we have a first party forum at Anaconda and it gets less and less use and also more and more robots. so we don't want to block the robots because the robots will inform your chat and so that's great but also how much time should I personally invest in responding to all of these answers on the forum if they've already been asked again well now I need to change my view of the value of this tool it is not just to get questions answered it is now to build how to build community and where is community being built I mean, not to call out Stack Overflow again, but they have now recently, I think it was 2025, they had the lowest usage of Stack Overflow and questions asked and answered since their inception in 2008. But that doesn't mean that a forum is no longer a good avenue. And so I hope they continue to build and exist as a question and answer platform. But now I'm looking at it to build community. And were the vanity metrics of yesterday going to be the vanity metrics of tomorrow? But also underscoring, they are all vanity metrics. It is high-level, easy-to-grab metrics that don't necessarily depict value. So maybe the user stories are the value. I don't know. So I'm putting my marketing hat on because these are questions that I think about every day of, like, how can me with only hopefully 35 hours, but probably between 35 hours and 60 hours a week, effect change and teach people and grow a technology and it's not easy uh so i i don't i don't have a good answer but that's that's my analysis on it

Speaker 2 [53:23]

Sounds like a good answer. I mean, I think what I take from that is, like, also, like, community, don't isolate, like, find other people, don't just sit and do this in the endless echo chamber. Maybe that's true for other aspects of life as well. A thing when I was learning to code that was always emphasized and I think is still very relevant is, like, reading other people's code. This, of course, becomes very interesting when other people's code is not written by other people but written by coding assistants. and I think that's going to be very curious to see how things develop. I was having a conversation with Kay earlier around, or I think it was Kay, around source code as artifacts, and it's a very curious point in time where we have all of this source code online, historically written by humans, and now we will move into an era where it's vastly going to be written probably by coding agents. So I think that's a very interesting technique, is see how others do it, be curious about this, get feedback, have other humans in the loop. I do think there's still plenty of ways to do that. So that would be my tip.

Speaker 3 [54:36]

I would like to say that it is totally fine for juniors to learn how to code by just using chatbots, but the job interviews are still requiring take-me-home challenges, which you can do with AI, but if it's obviously done with AI, it might be a problem. But you also can have live coding challenges and lead code and so on. So I think in the end, you need to find a way to still learn how to code. Yeah. So I like the example with learning from other people's coding example, and I think that's something that I did once when I was doing Advent of Code, where, like, that's, like, a really good source of people are writing their own code, because you're on a timeline, and you have to be fast, and you have to do it, and you have to do it with very few lines of code, and that was really cool, like, you do it, and then you look at the people on the leaderboard, how they did it, and you're like, wow, I didn't even know that library existed. That's a cool library. And it was very, so finding this cool coding challenges that are still forcing people to do software as a craft and not as a factory.

Speaker 4 [55:49]

One of the things that you both mentioned was the kind of reproducing the code, and I'll keep this really short, apologies, but the slop, the AI slopification potentially, and I do think, like, in the force multiplier, like, if someone released a very popular book on how to program, many of those have happened, then you will see a proliferation of that pattern out into the ecosystem. Well, now everyone has it at their fingertips. And if we continue to feed back in that pattern, and the pattern keeps coming out on the source, then we may build or bake in not particularly great patterns, and then when someone writes a very famous blog post debunking it, then we all kind of zoom towards this new solution. And so I just think it's a problem. I don't have a solution necessarily for it, but I think having the conversations, inspecting the validity. Why did you choose this source? What are alternatives? Can you critique your own code? What is another alternative? Give me two ways to do this one problem. And we're getting into prompt engineering here, which I used to look at incredulously, of like, who's this engineering? But I am a big tent engineer. All types of writing code is engineering. And I just think it's an interesting paradigm that I'm even, I'm growing into.

Speaker 1 [57:13]

So, yeah, we've come to the end of the session. I thank so much, and please join me in thanking our panel, Teresa, Don, and Jessica. I hope everyone, it was a very valuable session. I'm pretty sure everyone found it very, very helpful.

Speaker 2 [57:27]

helpful thanks for the great questions great yeah

Speaker 1 [57:30]

Yeah, indeed. So thanks a lot.

Speaker 3 [57:31]

Thanks a lot.