§ 01 Me
I'm Pranay, an AI harness engineer. I build reliable agent systems. AI amuses me.
I love researching how AI agents work, what makes them stand out from LLM wrappers. I deep-dive into OSS codebases of agent harnesses. Agent systems are a lot more than an LLM, a tool call, and a loop. The interesting work lives in the infrastructure built underneath: evals & tracing pipeline, communication channels, prompt compilation, tool policies, memory & compaction, callbacks & async abilities infra, etc. That is where I spend most of my time.
- Currently
- Learning
- Tooling
- Reading
§ 02 Learnings
Learnings gathered on agent infrastructure.
Detailed blogs on my learnings from building agent harnesses, reading OSS codebases like Codex, Claude Code, and openclaw, and inferring how the closed ones work from their leaked prompts. Why the systems around the harness matter as much as the harness itself. Why infra first, product second is the better design.