Sheet index
15 drawings · billed by weightGarage Connect
From on-device scraping to sales copilot, with a judge cutting forced pitches 100% to 8%.
Luca
CA assistant where the LLM emits a spreadsheet IR, never cell formulas.
EyesAI
An Android agent for blind users where only the Controller, never the model, declares success.
| Sheet | Title |
|---|---|
| 05 | HookRelayWebhook delivery that survives crashes, duplicates and receivers that stay down for days. |
| 06 | FS Preparation AgentFinancial statements where the LLM points at cells and arithmetic decides every link. |
| 07 | Saffron & SmokeVoice ordering where the agent can only speak prices its tools returned. |
| 08 | Agent Orchestration LayerTurned a single-user CLI agent runtime into a multi-tenant service without forking it. |
| 09 | GarageCrash-recoverable meeting note-taker that picks single-pass or map-reduce by transcript length. |
| 10 | JobTrackA Gemini insights layer that walks six models down when the free quota runs out. |
| 11 | SoulmateA paid matchmaking app where double-credits and double-charges are ruled out by schema constraints. |
| 12 | CityFixA civic issue-reporting app, 3rd Prize at GNEC 2024; I built the backend and schema. |
| 13 | SolidBoardA SolidJS Kanban board with two runtime dependencies and dense-integer card ordering. |
| 14 | Finance APIA typed-error REST API for expenses, with two dates per entry and a duplicate guard. |
| 15 | GameWebA browser gaming portal whose ETL merges two game feeds idempotently into one schema. |
Revisions
full résumé →Associate Software Developer (AI Engineering)
TechNsure IT Hub Pvt. Ltd. · Remote · Jan 2026 – Present
- A.1Measured 956 production recommendations, then replaced a useless score threshold with a judge that may say nothing fits: forced product pitches fell from 100% to 8%.
- A.2Built an agentic sales copilot over seven context layers that skips raw RAG when its relationship model is fresh, and reranks by ID only so it can't mislabel products.
- A.3Mirrored a 9-type catalog into Qdrant through a transactional outbox and HMAC webhooks, re-embedding only items whose text actually changed.
- A.4Moved Facebook and Instagram scraping onto users' devices, so no session is ever replayed from two IPs.
Bill of materials
79 parts · used in shipped workLanguagesqty 7
- Pythonshipped in FS Preparation Agent, Agent Orchestration Layer, Garage Connect
- TypeScriptshipped in Luca, Garage Connect, Garage, Agent Orchestration Layer
- Goshipped in DeployMachine
- Javashipped in HookRelay
- Kotlinshipped in EyesAI
- JavaScriptshipped in CityFix
- SQL
LLMs & agentsqty 19
- Agentic tool-callingshipped in Garage Connect, Agent Orchestration Layer, EyesAI
- LangGraphshipped in Garage Connect
- LangChain
- MCPshipped in Agent Orchestration Layer
- OpenAIshipped in Garage, Luca
- Anthropic Claudeshipped in Garage, Luca
- Geminishipped in Garage, JobTrack, Luca, EyesAI
- Vertex AIshipped in EyesAI
- Azure OpenAIshipped in FS Preparation Agent, Luca
- Groqshipped in EyesAI
- Perplexityshipped in Luca
- Ollama
- LoRA fine-tuning
- vLLMshipped in DeployMachine
- llama.cpp
- PyTorchshipped in EyesAI
- Hugging Faceshipped in EyesAI
- NumPyshipped in EyesAI
- Pandas
Retrieval & RAGqty 9
- RAGshipped in Luca, Garage Connect
- Hybrid searchshipped in Garage Connect, Saffron & Smoke
- RRFshipped in Saffron & Smoke
- HyDEshipped in Garage Connect
- LLM rerankshipped in Garage Connect
- Cohere Rerankshipped in FS Preparation Agent
- Qdrantshipped in Agent Orchestration Layer, Garage Connect
- pgvectorshipped in Luca, Saffron & Smoke
- FAISSshipped in EyesAI
Voice & realtimeqty 8
- Deepgramshipped in Garage, Luca, Garage Connect
- LiveKitshipped in Garage, Garage Connect
- Pipecatshipped in Saffron & Smoke
- Gemini Liveshipped in EyesAI, Saffron & Smoke
- Voxtral
- WebSocketsshipped in Garage Connect, Saffron & Smoke, Agent Orchestration Layer, EyesAI
- SSEshipped in Agent Orchestration Layer
- Silero VADshipped in Saffron & Smoke
Backendqty 12
- FastAPIshipped in EyesAI, FS Preparation Agent, Agent Orchestration Layer, Garage Connect, Saffron & Smoke
- Node.jsshipped in Garage, Garage Connect
- Expressshipped in Luca, Garage Connect, Garage
- NestJSshipped in Soulmate
- Spring Bootshipped in HookRelay
- RabbitMQshipped in HookRelay
- Celeryshipped in Agent Orchestration Layer
- BullMQshipped in Garage, Garage Connect
- Socket.ioshipped in Soulmate
- SQLAlchemyshipped in FS Preparation Agent, Agent Orchestration Layer, Saffron & Smoke
- GraphQL
- REST
Dataqty 9
- PostgreSQLshipped in FS Preparation Agent, Luca, Saffron & Smoke, Agent Orchestration Layer
- MariaDBshipped in HookRelay
- MongoDBshipped in Garage, Garage Connect
- Redisshipped in Garage, Garage Connect, Saffron & Smoke, Agent Orchestration Layer
- Drizzleshipped in Luca, Soulmate
- Prismashipped in CityFix
- Supabaseshipped in GameWeb
- Firebaseshipped in CityFix, JobTrack, Soulmate
- Neo4j
ML & visionqty 3
- TFLiteshipped in EyesAI
- ML Kitshipped in EyesAI
- Azure Document Intelligenceshipped in FS Preparation Agent
Cloud, infra & frontendqty 12
- GCP
- AWSshipped in DeployMachine
- Railwayshipped in CityFix, FS Preparation Agent
- DigitalOceanshipped in DeployMachine
- Dockershipped in FS Preparation Agent, Agent Orchestration Layer, HookRelay
- GitHub Actions
- Nginx
- Reactshipped in Luca, Saffron & Smoke
- Next.jsshipped in FS Preparation Agent, Saffron & Smoke
- SolidJSshipped in SolidBoard
- Linux
- TailwindCSSshipped in GameWeb, Saffron & Smoke, Soulmate
Recognition
all certificates →General notes
answers a recruiter usually needsWho is Mohammed Usmani?
Mohammed Usmani is an AI Engineer based in Mumbai, India, currently an Associate Software Developer (AI Engineering) at TechNsure IT Hub, working remotely. 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.
What AI systems has he actually built?
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. Garage Connect, an AI sales platform where an opportunity judge measured on 956 production recommendations cut forced product pitches from 100% to 8%, alongside an agentic copilot, a cited relationship model and on-device social capture. 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.
What is his experience with RAG and retrieval engineering?
He built a seven-layer prompt assembly for a sales copilot that cut estimated worst-case input from roughly 96,000 to 2,000 tokens per turn, with six Qdrant collections across the platform. The retrieval goes beyond nearest-neighbour search: contextual retrieval that prepends an LLM-written blurb before embedding, HyDE expansion for terse queries, a 14-day recency decay, LLM reranking, and reciprocal rank fusion of dense and text search in a webinar assistant. He has also worked with pgvector and FAISS.
Has he built realtime voice AI?
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.
What is the Cross-Reference Graph?
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.
Which programming languages and tools does he use?
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.
Is he available for hire, and where is he based?
Yes — he is open to AI/ML and backend engineering roles at product-focused startups. He is based in Mumbai, India, and can be reached at mohammedusmani2005@gmail.com or on LinkedIn.
Need someone to run your models?
I'm looking for AI inference and AI engineering roles. The fastest way to reach me is email; the résumé has everything on one page.