Full-stack engineer . I architect systems that respond in milliseconds, not seconds, and don't require 3 AM incident debugging.

• Building & Shipping
• Learning & Reading
• Writing & Sharing Knowledge
• Listening to something

• Core Tech — TypeScript, Next.js, React, Node.js, PostgreSQL, MongoDB, Redis, Prisma
• Extended Stack — pgvector, LangChain, tRPC, Inngest, WebSockets, Zod
• Infrastructure — Docker, AWS, RabbitMQ, Vercel, Git, Cloudinary

Kyron Durable PR review pipeline via Inngest. Tree-sitter AST chunking → pgvector retrieval → Gemini review → GitHub comment. End-to-end review under 45s for PRs < 500 lines. Incremental indexing cuts embedding API calls by ~80%.

CodeSense AI
RAG pipeline over GitHub repos. 768-dim pgvector embeddings with sub-20ms cosine similarity at 10k+ vectors. AST-aware LangChain chunking reduces prompt tokens by ~35%. tRPC for end-to-end type safety, Clerk for multi-tenant auth.

Kairo CLI
Agentic CLI on Gemini tool-calling. Autonomously chains up to 8 tool calls (web search, code execution) per task without per-step human confirmation. OAuth 2.0 Device Flow for headless auth. Session resumes in < 200ms.

Vartax
Real-time chat over Socket.IO with write-confirmed ACK. Message roundtrip under 80ms. 3-service decomposition via RabbitMQ; OTP event published and 202 returned under 5ms, decoupling 200–400ms SMTP latency.

• Open to SDE-1 / full-stack engineering roles
• Not deploying on Fridays
• Prototyping random ideas at 2 AM

Work · LinkedIn · GitHub · [email protected]

_{Always building. Always learning. Always shipping 🚀}