Hello everybody and welcome to the

GitHub podcast. We're in season two. We

missed you. My name is Kathy and I'm so

excited to be with you here today with

my new and shiny co-hosts Marlene and

GPS. Hey everybody.

>> Hello.

>> Hi.

>> Happy to be here and happy to be shiny

and new.

>> Yeah. I to be clear, I don't mean like

shiny skin or anything. I mean like

shine bright like a diamond.

[laughter]

>> I have oily skin. So, thanks for pulling

me out. No,

>> we could turn this into a whole skin

care podcast because I'd love to know

your routines. But anyway, hey

everybody. Um, I'm so excited to chat

with you both today and to kind of

introduce ourselves to the world what

this podcast is all about and also talk

about last season because I know with

the GitHub podcast last season you both

were in different parts of the

organization and a lot has changed in

the past 6 months and that's it's

mindblowing how much things have changed

and yeah, how have you been? What what

have you been working on? What did you

think of the last season? I really

enjoyed a couple of episodes of the last

season, but my favorite, and I think

this is going to be a sleeper pick, is

the Electron one

>> because

>> Okay, so I love hot takes, and there's a

bunch of hot takes sprinkled on like a

bunch of episodes last season, but the

hot take of how if anything's made in

like JavaScript or Electron or stuff

like that. It's going to be bloated.

It's going to be bad.

>> And Key, who's a core maintainer on the

Electron team, was talking about how

that's just not a

good take and talked a little bit more

about that. And as someone who does not

do web stuff as well, traditionally as

of recently, I think everyone's doing

web stuff now, right? It was nice to

hear from someone with experience and

who explained it in such like a cool and

kind way versus just like mean hot take

on social media.

>> JavaScript bad.

>> Yeah.

>> Yeah. [laughter]

>> It's too easy to dunk on. So, it was

nice to hear an experienced take.

>> Yeah. I liked that episode, too. And I

also have built Electron apps myself.

Webdev is my angle. And I always feel

like I'm defending against the dark arts

or rather defending against all of the

hot takes against it. But I get it. But

I agree with you. It's cool to hear

about the performance and how things do

well.

>> What about you, Marley?

>> Absolutely. I really like the episode

about MCP with Angie Jones. I thought

that was really good.

>> Yeah.

>> Um I generally am a big fan of MCP. It's

very controversial, I know, to this day.

I think I think like a couple of weeks

ago people said it was dead, but I think

it's not dead. I I feel like

everything's dead. Everything's dead in

software at this point.

>> It's like the new JavaScript.

>> It's the new [laughter]

>> the JavaScript of AI.

>> Yeah, exactly. Um, but I thought that

was really an interesting conversation.

and she talked a lot about Goose and

Henji I think is now actually heading up

the foundation that you know is is

stewarding MCP and so it's it was kind

of cool to hear her journey and her take

on that and yeah I generally am excited

about it as a technology so and I think

it was mentioned a couple of times last

season so yeah that was my favorite

>> yeah MCP model context protocol for

those who don't know we have a a couple

episodes on it last season and it's

glorified API calls but [laughter]

agent native

>> it's a lot more that's not

>> it's more THAN THAT YOU'RE

>> MORE THAN that it's a lot more than that

yeah

>> you're right you're right and that is

why we have two dedicated podcast

episodes to that

>> exactly

>> for me my favorite one was the tiny

winds episode that one is it's very near

and dear to my heart the tiny winds team

at GitHub because that team is all just

dedicated to fixing the paper cuts that

open source maintainers have to deal

with every single day where you don't

think about a lot of them. But when

you're really living on the platform and

building software on GitHub, there's a

lot of things that you have to navigate

all the time, whether it be aspects of

site performance or little UI changes

and stuff where one really small one

that I appreciate so much is for

contributing guidelines. Normally, you

have to look for the markdown file

that's on the main homepage of a

repository or in like a docs folder or

something and they just made it so it's

in the sidebar and so you can just click

contributing and it's right there no

matter where it is in your repo and I

love that. And anyway, they do a lot of

really cool things on the platform in

general. hearing them talk about it and

and hearing our guest Helen talk about

it when uh she was on the episode.

They're so passionate about it and they

really listen to users a lot and it's

very very maintainerdriven and and

specific to what maintainers want which

is what GitHub is all about.

>> Yeah, I knew someone was going to pick

that episode so that's when you

[laughter]

>> that was a good one

>> because it's a great it's a great one. I

think there's so much work that is done

that doesn't get the viral post, that

doesn't get the recognition,

>> but

>> matters a lot dayto day.

>> Mhm.

>> Again, a lot of times performance and

stuff, if it's

>> bad, everybody talks about it, but if

it's good, it's silent. It's it's just

hero.

>> Yeah. Unfortunately,

>> so funny as well. I remember I was

reading literally this weekend I read a

blog post about how like performance is

actually if you save time off of

performance you're actually saving a

human life because like all the time

someone would have saved like waiting

for stuff to boot up is actually like

someone's life you're saving so

life-changing work actually

>> real we'll need to drop that link

>> somewhere that share that article that I

will share it And uh yeah,

>> so

we should introduce ourselves a little

bit more because I I've been on the

podcast, so I'll do like a brief intro,

but also GPS, you've just mentioned that

you historically have not done any sort

of web development.

>> What do you do?

>> Well, I'm on a brand new team with you

and Marlene now. Prior to that, I was on

the Python team with Marlene. So for the

last year and a half or so

>> prior to that I was doing like

Linux cloud engineering op stuff. Prior

to that I was doing net stuff. Prior to

that I was doing CIS admin. Yeah I've

been I've been doing a lot but webdev

has never really been

um okay. I did I did um what's it

called?NET had uh Blazer. You ever heard

of Blazer?

>> Oh yeah. I've always wanted to try

Blazer.

>> Do you like it?

>> I mean, from when I used it years ago,

it was great because I could use .NET

and I didn't have to learn anything

else. But, you know, if you ask like a

actual webdev, who knows what their

opinion on it would would be.

>> Eh, opinions are just opinions.

>> True. But it it helped me accomplish the

task with the skill set that I had

>> and that was everything that I needed.

>> That's that's all you need. You also

mentioned cloud stuff. Can you speak

about learn to cloud a little bit?

>> Yes, I should speak about it more. Learn

to cloud. [laughter] When I joined

Microsoft 5 years ago,

>> I was like, I want to open source all

the knowledge that I have to kind of get

to where I'm at. So, I opened up a

GitHub repo and I was like, "This is

just going to be like a read me, nothing

else." So, I start typing my little Mark

now file and then after a couple hours,

I was like, "Oh my goodness, I'm going

to need something much more." [laughter]

>> Yeah.

>> So, then it it's evolved a couple of

times over the past 5 years, but now we

moved to a full Python stack back in the

beginning of this year, and we have

hands-on labs that are hosted on GitHub.

We have verification systems. We have

some AI stuff that's running to verify

people's hands-on progress. We have the

full curriculum, progress tracking. I

just wrote another phase that includes

everything that I know about interview

prep,

>> whiteboarding, about being able to speak

about yourself, having your own elevator

pitch, going to networking events,

everything that's helped me from like

the

>> get a job perspective.

>> So, I'm going to ship that

>> some point today. But yeah, it's an

open- source free educational platform

to learn the fundamentals of cloud

engineering. And we have like close to

7,000 people now using it.

>> That's amazing.

>> And yeah, it's it's the best thing I've

ever done in my career. And when I leave

tech, it'll be probably still the best

thing that that I did to help. Yeah. I

It's nice to get messages from people.

>> That's the joy of open source.

>> Yes. And I see people actually talking

about it online a bunch and being like,

"Thank you so much GPS because it's

super popular, which is cool."

>> Yeah, it's awesome.

>> Yeah.

>> Marlene, you were also on the Python

team. Tell us more about your journey to

here.

>> I have journeyied a long way with with

Gwen to be honest.

But um I've been at I've been at

Microsoft now for almost two and a half

years and uh have joined the team on the

Python team and have just been like in

the Python community for almost a decade

now I want to say. Um so before this

before joining Microsoft I worked um

I've I went through like a super long

journey of doing different things but

initially I worked as a software

developer at NVIDIA. So I worked on an

open-source framework. It's like a GPU

powered version of pandas called KDF.

And that was really cool because that

was one of my first experiences actually

like contributing to open source in a

major project. So I was helping kind of

maintain that project with some um they

have this was like the very early days

by the way of Nvidia when they were

still kind of coming out of the gaming

phase and just now getting popular for

AI stuff and so they didn't have a very

big Python developer community at the

time and they were trying to grow that.

So a lot of researchers were using um

the GPUs for data processing. So the

cyber was supposed to target that. That

was my first um job. And then I and and

during that and even before that I had

created like a nonprofit to teach girls

how to code in Python. I kind of was

transitioning from I had studied

molecular biology and then decided to

switch to computer science and was

studying that. Um, and during that time

I was like trying I had a small

nonprofit that was like teaching girls

how to code in Zimbabwe which is where

I'm from originally and really enjoyed

that. And [clears throat] part of that

process was really getting involved with

the Python community. So I was speaking

at a bunch of Pythons, volunteered for

the PSF, lots of different um activities

in the community, I would say, for a

couple of years and then I switched over

to developer advocacy because I was

interested in that and again was doing

kind of big data systems at uh a company

called Voltron Data and I was advocating

for an open source project called Ibis

and this is also like a pandas version.

uh it's it's a it it it's like pandas

but it's supposed to help with larger

data. So kind of like duct DB if you are

in that space but that was great. I

worked at that that was a startup for

about three and a half years and then I

moved over to Microsoft from from there.

>> Yeah.

>> That's so fun. And also don't forget all

the lang chain stuff too that you've

been working on.

>> Exactly. So yes, so once I moved over to

to Microsoft, I spent a good amount of

time uh actually when I was doing the

IVIS work, I did a bunch of lang chain

work then when it was very early days of

lang chain. And so when I joined

Microsoft, I was like, "Oh, I would love

to work on the Langchain stuff." And so

I got to actually start the uh Langchain

Azure repository and and actually

created that repository and have been a

Langchain champion community champion uh

for like for the 3 years I think now um

and helped contribute to that

repository. So yeah, I'm a big fan of

Lang Chain. Um have been advocating for

it inside Microsoft for a while now. So

yes, that's that's probably most of it.

I didn't realize your lang chain stuff

was even before Microsoft.

>> It was before this. Yeah. Yeah.

>> Early lang chain. That's like

>> early. They started it and then you were

like day two.

>> I was [laughter] absolutely I was 100%

there. And this is the thing. Anthony

actually who hired me to come and join.

I think he saw me tweeting about

Langchain when I was doing Ibis work

because

>> it was very close, super early I would

say to uh the agentic kind of workflows

cuz I was looking for this is when GBT

had just come out and I was looking for

a way to like process uh to have agents

talk to data that was in tabular data.

Um and Lang was super early to that.

Honestly, hats off to Lang Chain because

there's been a lot of different changes,

but they really started early and and

yeah, fortunately, I founded pretty

early as well, which is cool.

>> That's awesome. I love that you both I

love that we all have kind of ended up

in the same place but with such

different backgrounds because I feel

like that makes a team so rich with

information cuz you're you're coming

from again like GPS you're in like the

deep cloud space but all of the

different languages that end up in cloud

and then Marlene you're so deep in the

Python space that's so cool and then I'm

the web dev over here which is a very

different angle of everything where I

I'm a bit allergic to backend And um

I've done I' I've written Django before

so that's something right Django is

something

>> I like I liked Django about 12 years ago

when I used it but I come from a very

very webdev front end background where I

have been writing websites since I was a

very young teenager. I let me think when

yeah I was I was 13 but then also I I've

like dabbled in various other

technologies did Java first spell did

C++ first spell and then like swore off

of it for the rest of my life and and a

various other languages in between and

now I love my my front-end land where I

was early on in the React community. I

first started writing React in 2014 and

now I feel like React is like the web

language that all of the LLMs and and

tools are recommending people use

because it's just so big in the space.

Um, and so that's been fun to see and

and I'm very grateful to have that as a

background because it is so popular and

so deep in everybody's code bases right

now. Would you say that LLMs generate

decent React code?

>> Did the humans generate decent React

code before the LLM did? That's that's

>> that's what I'm asking.

>> Yeah, you know, sometimes I I'll leave

it at that. There's some things where

I'm just like, man, I know why it

generated that

as I delete it and and fix it because

and and then sometimes I'm just like,

oh, thank goodness it knows how to use a

use effect properly. It definitely as as

many developers say, it depends.

Sometimes

I'll leave it at that.

>> Yes. I also want to just mention that I

knew Cassidy even before joining

Microsoft because Cassie has like a mega

following online on Twitter. It's like

wild. I think it's like and there was a

time when you were I was seeing videos

of you like weekly because you're

>> putting out co right you were going hard

on videos I was alone and I was like I

might as well use my phone for something

[laughter]

I was making so many videos during co

times like and I start I had started

making like tech memes

>> more regularly in 2019 or so that was

something I'd always liked to do. I

gotta do a shout out to Tik Tok.

Honestly, I really liked making videos,

but it was such a pain to edit. And I

feel like we've hit such a renaissance

of video editing where all video editors

are way easier to use now. But before

Tik Tok,

>> the best video editor I really used was

Windows Movie Maker. Like it was

[laughter] this was so much effort to

ship a video. Um, and yeah, serious. I I

still have it installed on my computer.

I have like a really old version. But

anyway, Tik Tok really was just a good

avenue for getting some creativity out

there and then yeah, I was making silly

videos and then I had children and

[laughter] I lost a whole lot of time

for creative video juicing.

But I've been I've been following both

of you guys online so much, too, that

it's it's it's so exciting for me that

we're all on the same team and that we

can all be on this podcast together,

too, where we can talk about things that

our developers care about and our

communities care about and not just what

are the business objectives that we're

hitting in a given week uh at work. And

yeah, I think we're all I can speak for

us all when I say we're really excited

to be here and talk about things.

>> Yeah.

>> Yeah. Even though there's nothing more

fun than talking about business

objectives.

>> I ever since I was a kid, I have loved

turning unstructured data into business

insights.

>> Yes,

>> I just saw a meme about that. Like you

know how there's a meme when you're like

on a flight and someone will say like is

there a doctor here [laughter]

>> and someone else was like is there

someone who can share the uh KPIs with

the I don't know seuite or something

like I've got the OKRs. Yeah.

>> Yeah. [laughter]

>> Well, as much as I love talking about

business insights, I also want to talk

about our favorite open source projects

because what we do at the end of each of

these episodes is just talk about what

our picks are of a given week. And so

I'll start. The one that I wanted to

talk about is called Media Bunny. Um,

and Media Bunny is a JavaScript media

toolkit that is very, very cool. We will

drop a link in the chat in the show

notes where you can see it, but um it

lets you read, write, and convert media

with JavaScript. And so it's really nice

if you want to do some kind of like

video editing or tweaking or or changing

of images and stuff in your browser. And

it's very cool cuz it's all front end.

>> Not a sleeper pick. Maybe h

>> handy.computer.

>> So good. Yes,

>> I love dictating stuff. It's and it's so

simple, but it gets the job done so well

and it never fails. And I do a lot of

English, Spanish, Spanglish, because I

just that's the way I talk.

>> Yeah.

>> And seamlessly switching, no problem.

Knows what I'm trying to say. Very

rarely do I have to edit.

>> So good. I love it.

>> That's so good. Are you able to do

slang?

like like like like specific specific

multilingual slang. I've never tried

like could you say nanito or something?

I don't I've I've tried words with ena.

No problem.

>> Wow. We got to That's so cool. I haven't

tried

>> now. I need to test it to see where it

doesn't work.

>> Where are the edges?

>> Yeah.

>> Yeah. I I assume that when it's like

pretty heavy Spang English, it's

probably

>> gonna be struggle.

>> What is she saying?

>> Yeah. Like we're started in English, now

we're in Spanish. I think this word is

probably an word. Yeah, maybe. But

>> hi computer. Love it so much.

>> Computer very nice.

>> I will go next. So for my pick, it's try

kua or kua drivers. Um, and basically

these are like computer it's

infrastructure for it's open source

infrastructure for computer use agents

and it's very cool they have a

partnership with with Microsoft and and

um I've seen a bunch of example projects

people have built with it and basically

it's like taking um computer uses when

you're using LLM to use an app for you

or drive an app for you or maybe like go

to a website and click things and things

like that. So, uh I think they they have

some really really cool infrastructure

available if you want to use LMS with um

computer use. So, I've seen a a really

interesting project that they built

where it's kind of like using a sandbox

and then uh it's assisting people who

maybe can't type and then just need to

dictate but for very specific tasks um

in an app. So yeah, that would be my

pick right now.

>> That's interesting. Is it meant more for

local computers or like cloud computers

or both?

>> I think it's like ideally for local

first, but I think it works with both. I

I'm not really sure what the intention

was, but I think for local. That's what

I'm assuming.

>> Well, I will be diving into the show

notes for that one, too.

>> Yeah.

Well, all that being said, thank you

everybody for listening to episode 1 of

season 2 of the GitHub podcast. We have

loved chatting with you and we can't

wait to chat with you more. My name is

Cassidy. You can find me on their

internet as Cassadoo. C A S S I D O O on

most things. You could also Google

Cassidy Williams and there's me and a

Scooby-Doo character. And I am not the

Scooby-Doo character.

[laughter]

[gasps] And I am GPS. You can find me at

MadebyGPS

anywhere.

>> And I'm Marleene and you can find me at

marleene_zw

or at marleene.ai.

And yeah, I'm on most places. I'm mainly

on X, but most places.

>> And thanks for listening to the GitHub

podcast. This episode was produced and

edited by the team at Edit Audio. See

you next time. Bye.

>> Yeah. Bye

>> bye.