Why Quansight is Awesome!
I’ve been a big fan of the Open Source projects and the communities around them for some time now. A few weeks ago I shared my views and thoughts on the same in the post titled The Open Source Addiction. Gradually many companies have started embracing the idea of Open-Source Software. In fact there are a few companies supporting many of these Open Source communities. One such company is Quansight, with their Quansight Labs division aiming to sustain the future of Open Source.
In their own words,
Quansight Labs is a public-benefit division of Quansight created to provide a home for a “PyData Core Team” which consists of developers, community managers, designers, and documentation writers who build open-source technology around all aspects of the AI and Data Science workflow.
With this blog, I’ll try to share my experience as a summer intern at Quansight Labs. If you are interested in the technical details of my work, I’d recommend you to checkout the official blog post.
This was the first ever internship program hosted by Quansight Labs and I happened to be the very first intern. June 14 2021, my first day, it felt like a kid who had arrived in Disneyland for more than 3 months. I knew a lot of the PyTorch and PyData developers by github aliases already and was excited to finally meet them (virtually).
Since I was a given a choice to select one of the many interesting
intership projects, I already had a brief understanding about my work
and goals before day 1. From the very start, Ralf Gommers
(my mentor) helped me break down the end goal into smaller and more manageable goals.
With all the history, discussions and protocols attached to the space of
interoperability within array libraries (my project), I had a lot to catch up with
and I was given enough time to get comfortable before I made any real contributions.
With the help of Greg and
Peter I started making progress
on my first major goal. Both of them went out of their way to help me whenever I
had a hard time understanding something in
uarray. Even though Quansight is a
remote first company, they have a very friendly and welcoming environment,
and I always felt at home (excuse the pun).
Sometimes it can be hard to connect with people due to the remote first nature of the job. On the other hand, a flat hierarchy in the company makes everyone extremely reachable and accessible.
I made sure to setup small calls and meetings with all the amazing developers in Quansight. I planned at least two “getting to know/coffee” meetings every week for the next three months. In hindsight, this was the best thing that I did, interacting and sharing thoughts with all the developers helped me better understand the open-source space and broaden my perspective. With Donut, I enjoyed talking to someone new, matched randomly every week.
Throughout the course of the internship, I had the freedom to choose issues based on my interests within SciPy and PyTorch. Being closely connected with developers of SciPy, PyTorch team at Quansight and Facebook (Thanks Mike) helped a lot when making that decision.
All of this already speaks a tonne about the atmosphere and the collaborative culture at Quansight, but I’d like to highlight a bit more.
As you would expect in an ideal company, they have an open and transparent system which is deeply rooted in the company values. It’s just one of the things that the team is doing by creating an environment that is appealing to everyone. It is hard for startups and companies in the open-source space, to directly compete with big tech MNCs only on compensation. Quansight still manages to attract super talented people who could also choose to work at these big MNCs given their skills and experience, but, they don’t.
In a casual conversations with Ralf, he said
What we can do is aim to offer:
- Interesting projects with quite a bit of freedom.
- Friendly colleagues.
- Not much work pressure or crazy work hours.
- Transparency in how we all work and how the business works - this also teaches skills beyond pure tech skills.
In my short three month stay, I can only say that the Quansight team absolutely nails each of those points. Quansight folks and in general people within the open-source communities are not motivated by the monetary benefits (Don’t get me wrong, I’m sure you’ll be paid competitively here), they are just really passionate about their work, learning, and giving back to the community.
You will find all kinds of problems to engage yourself in. Some projects might require a breadth of knowledge, while some might need you to dive deeper and focus on a niche field requiring you to learn low level concepts. This is a great way to explore different domains and also develop an expertise. There are several people in the team working towards accessibility with initiatives bringing the traditionally marginalized voices of disabled scientists into scientific computing communities via building and applying accessibility tools, standards, and community contribution practices in projects like JupyterLab, NumPy etc.
All in all you can expect working on some projects of your interest, making big contributions and real impact. But, it is not only about working individually, an open source community is powerful due to its decentralized development system and a collaborative environment. People at Quansight are always looking to bounce ideas off each other and share interesting thoughts.
Tech shares (aka ‘tqz’) are organized every week, where we generally talk about anything and everything ‘tech’ that someone encountered over the week. Discussions ranged from developer tools to programming demos to smart home devices and everything in between. The intern cohort had a separate weekly internship share which was based on a similar concept but was specifically run for the interns only to make us feel comfortable and provide us with a platform to talk about our projects. Shout out to Melissa for organizing these internship shares. These were a great opportunity to learn and interact with the team. PyTorch team had their own specific version called “PyTorch Tech Share” which I absolutely adore. These shares were so so informative and helpful for someone who would like to contribute to PyTorch. The discussions in these meetings are something that you will not find in a blog post, a book or a youtube video making them very unique.
I hold a major in Earth Sciences but through my experiences and interest in Machine Learning, I ended up doing a lot of Computer Science specific work. In the past, during my undergrad sophomore days, I used to question my abilities since I was not a so called “Computer Scientist” by degree.
It was interesting to see that not all (almost 50%) team members in Quansight hold a CS degree. You will find people with a masters or a PhD. in Math, Physics, Stats, Biology, Geology and other STEM fields. There are different ways to learn and all of them don’t need a degree. Folks at Quansight understand this fact and unlike some other companies they don’t set a hard degree requirement for any job position. I guess the only requirement is a genuine interest in open source and a desire to learn and upskill.
Even though this was just the very first year of the internship program, I felt everything was smooth and nicely structured. The project was clearly defined, still offering a lot of flexibility and freedom to choose from
Thanks to Ralf Gommers (who is easily the best mentor ever) for always being extremely supportive. Ralf is someone who almost always prioritizes mentoring/helping others over his own deliverables, and that also scales much better for the whole team/company. Working directly with someone who you respect and think highly of is always a great experience. I always felt more confident after our discussions and 1-on-1s every week. You inspire everyone at Quansight and the broader PyData ecosystem. Thank you for everything!
It’d be unfair to end this blog without mentioning Kush who was someone I always went to with random doubts and questions around PyTorch, Python, Open-Source, DSA and so many more topics. Thanks Kush!
Last 3-4 months have been so much fun, time flied and here I’m reminiscing summer of 2021. It is hard to put it in words but as I said earlier working at Quansight Labs has been nothing short of a Disneyland dream which is finally real. Quansight is just an awesome place to be at for any open source developer.❤
People at Quansight are doing a great job in the PyData ecosystem and several other open-source projects, but it’s a shame that we don’t talk enough about it. I feel we should highlight their work since most of these developers are extremely modest and don’t really talk about their contributions in the community. I’ll leave you with a few links to checkout. Highly recommend watching this amazing podcast by Lex Fridman featuring Travis Oliphant who is the person behin Quansight, Anaconda, NumFocus, SciPy and NumPy. Believe me, these will be the best 3 hours of your week, if you are interested in the PyData open-source world. My friend Gagan wrote a great post sharing his first month experience at Quansight. If you are looking for something more informal, watch this live discussion he had with Kush on youtube.
Until next time!