Systems Optimization Copilot

Inspiration

I’ve lost countless hours to CUDA build errors, linker failures, and cryptic compiler output. I wanted a tool that could read an error log, search relevant docs and forum threads, and explain why the error happens along with concrete fixes.

What it does

Systems Optimization Copilot is an AI debugging assistant for CUDA, C++, linker, and build errors. You paste an error log, and it runs a multi‑step pipeline: search → filter results → extract content → reason → format. It returns a root‑cause analysis, step‑by‑step fix instructions, and sources.

How we built it

Backend: FastAPI with an async pipeline and SSE progress updates. Integrations: You.com Search + Contents + Agents APIs for grounded research and reasoning. Frontend: Next.js + Tailwind with a terminal‑style UI and live “thinking” steps.

Challenges we ran into

Handling different API schemas and response formats. Streaming progress to the UI while the agent reasons. Making the output structured and actionable, not just a generic response.

Accomplishments that we're proud of

What we learned

How to build reliable, multi‑stage AI workflows. How to ground reasoning in live sources and present clear fixes. How to design UX for trust and transparency in AI debugging tools.

What's next for Systems Optimization Copilot

Built With

• agents)
• contents
• events)
• frontend)
• httpx
• python-(fastapi)-next.js-+-react-tailwind-css-you.com-apis-(search
• render
• server?sent
• sse
• vercel