I build AI agents that ship real code, and I own the review of everything they produce.
The evidence: I wrote an agent that finds actionable issues in open-source repos, writes the patch, and opens the PR. 17 of those are merged into third-party projects - langflow, dify, Grafana, Flowise, twenty, agno, Directus, pnpm, Medusa, Haystack, windmill, Effect, logto, reth, supabase-js, graph-node and Temporal's TypeScript SDK. Around 726,000 stars combined, all merged between March and July 2026.
Getting a patch past a maintainer who has no reason to be generous to you is the hard part, and it is the part I would point at. The agent does not fail at writing code - it fails at judgment: which issue is actually fixable, when a green test suite is lying, when a one-line fix is really an API change.
Also built Agent Vault, a runtime safety layer for AI agents (Rust, gRPC, NATS) with SHA-256 state verification and a four-tier escalation policy. Day job is product engineering; 15+ freelance projects delivered end to end; CodeChef global rank #22.
Looking for part-time contract work - roughly 15-20 hrs/week, IST with several hours of European or US-East overlap. Most interested in agent infrastructure, evals and developer tooling, but good full-stack product work is welcome too.
ayushbaluni · · focus · HN ↗
Remote: Yes - remote only
Willing to relocate: No
Technologies: TypeScript, React, Next.js, Node.js, Fastify, Python, Rust, PostgreSQL, MongoDB, Redis, Docker. Agent work: Claude Code, Cursor, MCP, Anthropic + OpenAI APIs, agent orchestration, evaluation harnesses. Also Shopify (Python connectors, inventory and placement CRUD), Solidity/Solana.
Resume: <a href="https://www.ayushbaluni.in" rel="nofollow">https://www.ayushbaluni.in GitHub: <a href="https://github.com/aayushbaluni" rel="nofollow">https://github.com/aayushbaluni LinkedIn: <a href="https://linkedin.com/in/ayush-baluni" rel="nofollow">https://linkedin.com/in/ayush-baluni Email: abaluni6@gmail.com
I build AI agents that ship real code, and I own the review of everything they produce.
The evidence: I wrote an agent that finds actionable issues in open-source repos, writes the patch, and opens the PR. 17 of those are merged into third-party projects - langflow, dify, Grafana, Flowise, twenty, agno, Directus, pnpm, Medusa, Haystack, windmill, Effect, logto, reth, supabase-js, graph-node and Temporal's TypeScript SDK. Around 726,000 stars combined, all merged between March and July 2026.
Getting a patch past a maintainer who has no reason to be generous to you is the hard part, and it is the part I would point at. The agent does not fail at writing code - it fails at judgment: which issue is actually fixable, when a green test suite is lying, when a one-line fix is really an API change.
Also built Agent Vault, a runtime safety layer for AI agents (Rust, gRPC, NATS) with SHA-256 state verification and a four-tier escalation policy. Day job is product engineering; 15+ freelance projects delivered end to end; CodeChef global rank #22.
Looking for part-time contract work - roughly 15-20 hrs/week, IST with several hours of European or US-East overlap. Most interested in agent infrastructure, evals and developer tooling, but good full-stack product work is welcome too.