[GSoC 2026] Interest: Chipathon Knowledge Hub + "Ask Chipathon" Chatbot

[harsh-kumar-patwa]

Hi @msaligane and @luarss,

I am reaching out to express my interest in the Chipathon Knowledge Hub + "Ask Chipathon" Chatbot project for GSoC 2026.

What really draws me to this project is the problem it is solving. Chipathon participants deal with scattered information across GitHub repos, Slack threads, Google Docs, and past guidance — and that makes onboarding and debugging unnecessarily painful. Having a single knowledge hub backed by a RAG chatbot that can point users to the right answer with citations feels like it could genuinely change the participant experience.

I am a Computer Science student at BITS Pilani, currently interning as an SDE. I have a strong interest in AI/ML and have hands-on
experience building RAG-based systems — including one that implements a full pipeline with embedding, vector indexing, query
classification, retrieval, and response generation with source attribution using Gemini API. The architecture is very similar to what the "Ask Chipathon" chatbot would need — classifying user queries, retrieving from domain-specific documents, and generating grounded responses with clear citations. Applying this to a domain as specialized as chip design is genuinely exciting to me.

A few things I have been thinking about for this project:

• Source prioritization — What would be the most valuable sources to index first? I imagine past Chipathon debugging threads and OpenROAD flow logs/artifacts (DEF, GDS, metrics) would be high priority, but I would love to hear what the mentors think.
• Evaluation harness — The deliverables mention 30-60 Chipathon-style questions for regression checks. Are there existing FAQ collections or common questions from past Chipathons that could seed this?
• Maintainability — Since the knowledge base needs to stay useful across Chipathon seasons, I am curious about how the team envisions the contributor workflow for mentors to add or update answers during the event.

I would love to start working on a small proof-of-concept if there are any specific data sources or documentation you would recommend I look at first.

Looking forward to hearing from you!

Best regards,
Harsh Kumar
LinkedIn | GitHub

[luarss]

Hi @harsh-kumar-patwa ,

Thanks for your interest! Please stay tuned for the finalized project list, which will be available in early March 2026.

Best,
The OpenROAD Initiative

[luarss]

Hi @harsh-kumar-patwa , please check out the following link for the finalised projects. We have also created a discord for this year's edition, do hop on and say hi!

Project List: https://docs.google.com/document/d/1X6xxUonxgEQ_iD5G5vFp2ZOH1vbkRSspEdYrpSf4khE/edit?usp=sharing
Discord Link: https://discord.gg/axCZmUbH

[musaabhasan]

For an “Ask Chipathon” knowledge hub, I would make source governance the main design feature rather than treating it as a normal chatbot wrapper. The hard part is not only retrieving content; it is preventing stale or unofficial guidance from sounding authoritative.

A practical MVP could use:

• source registry: GitHub docs, official workshop material, issue threads, accepted PRs, and curated FAQs with owner and last-reviewed date
• ingestion policy: exclude private Slack excerpts unless explicitly approved and summarized into public/approved notes
• retrieval filters: project phase, toolchain area, version, and source authority
• citation rules: answer only with source links and clearly mark “no authoritative source found”
• evaluation set: common onboarding, build failure, flow debugging, and submission questions with expected source IDs

That gives mentors something auditable: when the bot answers, they can see which source was used, whether it is current, and whether the answer is grounded in official Chipathon material.

[luarss]

@musaabhasan Thank you for your interest. However, applications for GSoC 2026 have concluded, and we are no longer accepting any further applications.