Software Engineer building high-quality AI applications, agentic systems, and AI memory & context infrastructure.

I've spent 10+ years taking products from idea to production across AI, fintech, and data platforms — currently building generative voice infrastructure at Resemble AI, where I own model serving and training pipelines for production TTS. Along the way I've shipped full products end to end (frontend, backend, DevOps), led teams, and open-sourced tools used by thousands of developers.

These days my focus is making AI systems that are reliable, fast, and cost-efficient — agentic workflows that actually ship, and memory/context layers that let agents reason over large codebases and data without drowning in tokens.

See the flow of your spaghetti code.

An AST-based knowledge graph that gives coding agents architectural memory of a repository — tracing semantic call paths across microservices and abstraction layers so an agent can localize a feature or bug without reading the whole codebase.

• ⚡ Zero-token extraction — architecture is mapped from the AST, not the LLM
• 🎯 62.5% Recall@1 on SWE-bench Lite issue localization
• 💸 ~2,400–3,200 tokens to answer, vs. 8,200–28,500 for comparable approaches
• 🤝 Agent-agnostic — works with Claude, Cursor, Gemini, and others
• 🔁 SHA-256 incremental updates + a shared, version-controlled architecture map

End-to-end AI-led screening, interviews, and evaluation that cuts hiring cycles from weeks to days. Automated resume screening, AI-conducted interviews, coding assessments, and adaptive AI-fraud/proctoring detection with full audit trails.

Turns natural-language ideas into working, deployed software. Autonomously generates the UI, builds Docker images, provisions Postgres, and wires in AI capabilities (OCR, speech-to-text/TTS, real-time voice, LLM business logic) — with a human-in-the-loop engineering safety net for production reliability.

AI / LLM  Agentic systems · AI memory & context engineering · PyTorch · Model serving (TTS / Generative) · Latency optimization · Model routing

Backend  Python · Node · Django · GraphQL

Frontend  React · Angular · TypeScript

Data  Spark · Elasticsearch · Redis · Druid · Celery

Cloud & DevOps  Docker · Kubernetes · Helm · AWS · GCP · Azure · Jenkins · CI/CD

 📍 Bangalore, India

_{⭐ If you're building agents that need memory, check out TLDRGraph — feedback and stars welcome.}