Mohammed Usmani · AI EngineerI build agentic AI that runs in production.

Mohammed Usmani is an AI Engineer based in Bengaluru, India, currently at Garage (Gravitichain Technology Group). He builds production agentic AI systems — multi-agent orchestration, retrieval pipelines, realtime voice, and the multi-tenant backends behind them — and has shipped 15+ of them across sales, accounting and workspace products.

Four stand out. A financial-statement engine whose Cross-Reference Graph links every statement line to its note and ledger account, confirmed by arithmetic tie-out rather than model judgement. NetworkChains, an AI sales platform combining an agentic copilot, a realtime call assistant, a relationship graph and conversational image editing. Luca, a ~199,000-line AI accounting platform for Indian Chartered Accountants running on five LLM providers. And EyesAI, an autonomous Android agent that operates a phone end-to-end for blind and low-vision users.

He built a seven-layer prompt assembly over six purpose-scoped Qdrant collections that cut worst-case input from roughly 96,000 to 2,000 tokens per turn. The retrieval goes well beyond nearest-neighbour search: contextual retrieval that prepends an LLM-written blurb before embedding, HyDE expansion for terse queries, reciprocal rank fusion with a 14-day recency half-life, hybrid dense and lexical search, and LLM reranking with diversification. He has also worked with pgvector, FAISS and Pinecone.

Yes. He built a live in-meeting copilot that captures the host microphone and remote participants as two separate streams, downsamples 48kHz to 16kHz PCM16 in 250-millisecond batches, and gives each stream its own Deepgram socket — so speaker attribution comes from the transport rather than a diarization model. He also built a voice-to-voice restaurant ordering system on Pipecat and Gemini Live over a full-duplex WebSocket, and telephony voice agents over Telnyx SIP.

It is the core data structure of the financial-statement engine: 9 node kinds and 6 edge kinds modelling how a set of financial statements actually hangs together. Links are made by deterministic resolvers and then confirmed by arithmetic tie-out, and every edge records how it was created and whether the numbers reconcile. Where an LLM assists, it only extracts and points at cells — the pass/fail verdict comes from the same deterministic arithmetic, so no reconciliation passes without a verified tie-out.

Primarily Python with FastAPI and TypeScript with Node, Express and NestJS. Day to day that means LangGraph, the Model Context Protocol, OpenAI, Anthropic Claude and Gemini, Deepgram and Voxtral for speech, Qdrant and pgvector for retrieval, Celery and BullMQ for background work, PostgreSQL, MongoDB and Redis for storage, and Docker with GitHub Actions deploying to GCP, AWS and DigitalOcean.

Yes — he is open to AI/ML and backend engineering roles at product-focused startups. He is based in Bengaluru, India, and can be reached at mohammedusmani2005@gmail.com or on LinkedIn.