fizza_pizza

This is actually a really fun idea. I’d honestly trust routes made by real people way more than another “top 10 places” blog post.

This is really cool. “I wanted an excuse for writing Clojure while pretending to write Go” genuinely made me laugh. The 7ms startup time and Plan9 support are both kind of wild too.

This actually makes a lot of sense. Feels like most people are trying to brute force reliability with bigger models while you’re reducing the problem space instead. “Agents are suggestions, states are laws” is such a good line too.

This is seriously impressive. A single pass LLVM frontend in ~3k lines of C with no malloc or AST is kind of wild. The graphical examples were a really nice touch too. Curious to see how far you can push the IR.

The "no FFN for retrieval tasks" observation is interesting essentially arguing that if the knowledge is in the context, FFN weights are redundant for that task. Curious whether this generalizes to multi-turn tool calling where the model needs to track state across several calls, or if it breaks down there. Single-shot is the easy case.

The certification angle is the most interesting part to me. Regulated industries (aviation, medical devices) often can't use JIT for exactly this reason, the code that runs has to be the code that was certified. Static translation that produces a signable binary is a real unlock there, code bloat notwithstanding.

Location: Karachi, Pakistan

Remote: Yes

Willing to relocate: Yes

Technologies: Python, TypeScript, React, Next.js, FastAPI, Node.js, AI Agents, LLM Applications, Docker, Git/GitHub

Résumé/CV:https://drive.google.com/file/d/1gCqD10dfO6tMf_HjsOE8KCUpWag...

Email: [email protected]

Looking for remote internships, freelance projects, or full-time opportunities in AI, backend, or full-stack development.