I design and ship AI systems that hold up in production - not just in a demo. Voice AI · Agentic architectures · Multi Agents · MCP .
| Products shipped & sold | Legal AI → acquired by a US law firm & VALSCO · Text-to-SQL engine → acquired by a Bangalore startup |
| Currently building | MemDream — open-source memory layer for AI agents |
| Freelance / private work | 20+ delivered projects |
| Open source | 1000+ contributions in the last year |
| Research | 4 papers · 2 patents filed |
| Teaching & speaking | 10 internship programs led · 9 guest talks · 1000+ students mentored |
Voice AI - real-time STT/TTS pipelines, sub-second transcription, interrupt & turn-taking handling, telephony and WhatsApp Calling, avatar-based and on-prem voice agents.
Deepgram Whisper LiveKit WebRTC Twilio
Agentic Systems - multi-agent orchestration, A2A communication, tool calling, agent memory, structured outputs, long-running workflows. Shipped a system with 7 agents, 52 tools, full A2A comms.
Backend & Distributed Systems - gRPC, SSE, WebSockets, event-driven pipelines, async processing at scale.
FastAPI Go Node.js Kafka RabbitMQ Redis
Infrastructure - containerized, orchestrated, observable systems.
Docker Kubernetes Terraform Azure GCP
My path started in robotics - autonomous rovers, YOLO-based vision, edge inference on Jetson/TensorRT — before moving into large-scale AI systems in production.
