[Nomination] Herumb Shandilya #98
Description
Select one:
- I am nominating myself for the PyTorch Ambassador Program.
- I am nominating someone else to become a PyTorch Ambassador.
Please confirm that the nominee meets the following requirements:
- The nominee is 18 years of age or older.
- The nominee agrees to abide by the PyTorch Code of Conduct.
- The nominee agrees to comply with the Linux Foundation Antitrust Policy.
- The nominee meets at least one of the qualifications listed in the PyTorch Ambassador Program Requirements.
Nominee Name
Herumb Shandilya
Nominee Email
Nominee's GitHub or GitLab Handle
https://github.com/krypticmouse
(Optional) Organization / Affiliation
Stanford University
City, State/Province, Country
Stanford, California, United States of America
Your Name
No response
Your Email (Optional)
No response
How has the nominee contributed to PyTorch?
- An active contributor to PyTorch repositories (e.g., commits, PRs, discussions).
- A speaker at PyTorch events or workshops.
- A PyTorch user group organizer or meetup host.
- A researcher or educator using PyTorch in academic work or training.
- An active leader in the PyTorch community with at least one year of experience in:
- Organizing events (virtual/in-person).
- Speaking at AI/ML conferences.
- Mentoring others in PyTorch.
- Creating technical content (e.g., blogs, videos, tutorials).
🏆 How Would the Nominee Contribute as an Ambassador?
As someone who has organized cohorts on PyTorch centric usecases in Cohere for AI for 1.5 years, I've come to realize that teaching how to use something promotes projects with great potential but if you tell some how that something works you are building a community of contributors then.
Pytorch is a great frameworks and part of the reason is that we have "enough" level of control. However, the core components that power it are usually behind the scenes and often seen as "exotic". I love to to make these more accessible and normal to the user.
To be more concrete, I would love to conduct project sprints and develop content that demystifies these concepts in a more grounded approach i.e. if you were to discover this thing from nowhere how would you do it? The more approachable a code-base seems the more comfortable new contributors will feel.
Logistics wise, I can organize learning groups at Stanford too, we have a big userbase of PyTorch users here!Any additional details you'd like to share?
I've been writing blogs and conducting educational cohorts for ~5 years now. It was in 2020 that I along with some passionate folks started "Build With Us" Cohorts at Crework. Here we chose a small set of peers divide them into teams and helped them build real-world products rather than simple PoC projects. We had weekly reading and implementation sessions on various topics revolving around various topics but implementations centering around PyTorch.
In 2023, I started a channel "Beginners in Research Driven Studies"(BIRDS) at Cohere for AI, along with my 2 co-leads where we helped beginners to understand the process of research. Over the period of my time there we offered 3 programs:
1. Mini Sprints: We choose a Paper to read and in week 1. We choose a lead to run the sprint who would explain the project from first principle and explain the paper. In week 2, that person would show case the implementation of the project which would be in PyTorch. In week 3 we had a guest speaker(usually the author of the paper) and in final week we showcased implementations of people in the community.
2. Nidification: These we one-shot sessions centering around explanation of niched concepts.
3. Mini Cohorts: This was an 8 week event with a more zero-to-hero agenda. I'll give the example of CUDA Cohort that I lead while I was there. We started by creating a syllabus for the 8 weeks without assumptions or pre-requisites along with weekly assignments for practice. Each week, we would give a session in a lite lecture type setting explaining concepts like GPU internals, C/C++, Working of various CPU components, Parallel Programming, PyTorch Internals, CUDA C, etc. Last 2-3 weeks, we would have a capstone projects that people would make. In our first cohort we had a registration of more than 150 people and many of them had really great Ideas to work on.
I love communities and building them, they are not just a learning opportunity but a group a people with same goal. That is a motivation in itself, talking to people smarter than me is a privilege that becomes more accessible in communities.
Blogs: https://journal.herumbshandilya.com/, https://www.geeksforgeeks.org/user/herumbshandilya/contributions/
Beginners in Research Driven Studies: https://sites.google.com/cohere.com/coherelabs-community/community-programs/birds