-
🔭 I’m currently working on Nexonomy, a multi-agent platform that plans, builds and deploys full apps from one prompt
-
🏢 I’m rolling it out at a Fortune 500 company as a decision-intelligence platform over their own data
-
🧠 I work with LangGraph, LangChain, RAG, agent memory, evals and self-hosted open-weight models
-
🌱 I’m currently exploring MCP for coding agents and running open-weight models on our own GPUs
-
💼 5 years of full stack work, 2+ years building production LLM systems
-
💬 Ask me about multi-agent orchestration, RAG pipelines, Node.js, React and React Native
-
📫 Reach me on abdulmoeez0812@gmail.com
-
⚡ About Me I thrive under pressure and love to work on challenging projects
|
🤖 Nexonomy Lead engineer on a multi-agent platform. A supervisor agent routes work across planning and build pipelines, role-based agents debate and replan in parallel, and humans approve each step. Backend, frontend and integration agents write the app, error-resolver agents fix it in E2B sandboxes, and Playwright and Browser Use vision agents test it before it ships to its own live subdomain. LangGraph · LangChain · LangSmith · FastAPI · Node.js · TypeScript · MongoDB · Redis · Next.js |
🏢 Enterprise rollout, Fortune 500 Extending Nexonomy into a decision-intelligence platform with governed answers over company data. Code computes every figure, a triage model routes each question, and advisor and critic models review each draft. Includes ingestion and RAG with PII redaction and a knowledge graph, agentic RBAC with SSO, long-term memory and golden-set evals. RAG · pgvector · Evals · Kubernetes · Helm · ArgoCD |
|
🖥️ Open-weight models Evaluated and self-hosted open-weight LLMs on L40S and A10G GPU nodes, with enforced structured output, tool calling and per-workflow model switching, for workloads that can't leave the company network. Self-hosted LLMs · GPU serving · Model routing |
🩺 MedWrite Built the mobile app and the first AI layer of a clinical documentation platform that turns clinical input into structured medical notes. LLMs · React Native · Prompt design |
|
🧵 Threads Core engineer on a multi-tenant BPM platform: Form Builder, drag-and-drop Workflow Builder, Automator and Instantiator. Built the workflow engine with role-based access, retryable queued jobs and BullMQ audit logging. React · Redux Toolkit · Node.js · MongoDB · BullMQ · Turborepo |
🎓 ResparLink Full stack on a medical research mentorship platform (web and iOS). Built an AI helper agent for researchers on LangGraph, per-role data isolation with Supabase RLS, and an xAPI LRS integration on AWS ECS. LangGraph · Supabase · PostgreSQL · AWS ECS |
|
💳 Krillpay & Chip Tech lead on a multi-currency payments platform with Cybrid, Persona and Plaid and idempotent, retry-safe transaction flows. Led the rebuild of the Chip e-wallet from native Swift and Kotlin into one React Native app, with Facia liveness KYC. Node.js · React Native · Payments · KYC |
🧘 Free Hypnosis & reverse image search Owned delivery on a React Native wellness app with 1M+ downloads and built subscriptions that grew revenue by about 50%. Earlier, built a reverse image search engine on an AlexNet CNN, from data pipelines to REST APIs. React Native · Python · Computer vision |
- Agents that can be trusted with real data: code computes the numbers, models explain them, and a critic model reviews every answer before a person sees it.
- Humans stay in the loop: approvals, audit logs and role-based access are part of the design from day one, not added later.
- Evals before opinions: golden sets and regression evals decide which model or prompt ships.
- Full stack ownership: from the LangGraph graph and the API to the React screen and the Kubernetes deploy.
