Hasnat A.

Three AI products with real users, shipped end to end: OmniLegislation (AI legal monitoring across all 50 US states), Sense (18M+ social impressions analyzed for brands including Westfield London) and Oracio (voice-to-document AI SaaS in 52 languages). I build the same way for clients: RAG pipelines, AI agents, voice agents, n8n automations and complete AI SaaS products, deployed and monitored, not demoed. 6+ years full-stack. Top Rated on Upwork, 100% Job Success, 1000+ tracked hours. Anthropic certified (Claude 101, Claude Code in Action). WHAT CLIENTS SAY Recent Upwork work includes a production document analyzer built with Claude and a NestJS/Node.js API layer restructured for a US product team: "The developer clearly knows what they're doing when it comes to AI systems and delivered a solid, production-ready document analyzer." "Hasnat was a great find. He restructured the API layer so it's far cleaner and easier to work with." IF ANY OF THESE SOUND FAMILIAR, I CAN HELP - "Our chatbot makes things up." - "The agent works in the demo and breaks on real data." - "Retrieval keeps missing the right documents." - "Our OpenAI bill is out of control." - "We need an MVP with AI in it, live in weeks, not months." WHAT I BUILD - AI agents for business automation: tool calling, MCP, human-in-the-loop approval, LangGraph - Voice AI agents: inbound and outbound phone agents that qualify callers, book appointments and update your CRM, with fallbacks and call analytics - Workflow automation with n8n and LangGraph: CRM enrichment, document routing, reporting, approval steps - Custom MCP servers so Claude can read and write inside your CRM, database or internal tools, with scoped permissions and every call logged - RAG systems over PDFs, websites, Notion, Confluence and databases, with hybrid retrieval, reranking and an evaluation set - AI customer support chatbots and internal copilots with source citations and handoff to a human - Complete AI SaaS platforms: multi-tenant, billing, roles and permissions, admin dashboards, deployed on AWS - Full-stack MVPs (web and mobile) built to survive their first thousand users HOW I WORK Every system ships with evals, fallbacks, cost controls and monitoring, because an assistant that is right 80% of the time is a liability, not a product. I start from the business bottleneck and the data you actually have, not from a framework. You get one person who can take the executive conversation, the architecture, the code and the rollout, without handoffs. GOOD FIT IF - You have a real process or product to improve and a budget of $1,500+ for a build, or $500+ a month for a managed agent - Someone on your side can make decisions and grant access NOT A FIT IF - Price is your first filter - You want a demo rather than a system Domains: legal and compliance, SaaS, marketing and social analytics, meeting and document intelligence, regulated workflows where audit trails matter. STACK AI: Claude, OpenAI, Gemini, LangChain, LangGraph, LlamaIndex, MCP, Pinecone, Qdrant, pgvector, Chroma Voice: Twilio, Deepgram, ElevenLabs, OpenAI Realtime Automation: n8n, Make, LangGraph Data: Python ETL/ELT, Airflow, dbt, BigQuery, Snowflake, PostgreSQL, Redis Backend: Python (FastAPI, Django), Node.js (NestJS) Frontend: React, Next.js, TypeScript, Tailwind Infrastructure: AWS, Docker, Kubernetes, Terraform, CI/CD Tooling: Claude Code, Cursor, Lovable, Replit, MCP servers Message me with what you are trying to automate and where it is stuck. I will reply with a short walkthrough of how I would build it, and whether AI is the right tool at all, before you spend anything. Keywords: AI Engineer, AI Agent Developer, Voice AI Agent Developer, Conversational AI, RAG Developer, LLM Developer, AI Chatbot Developer, n8n Automation, AI Automation, AI Integration, Generative AI, Claude, OpenAI, LangChain, LangGraph, MCP, Vector Database, AI SaaS, MVP Development, Full-Stack Developer, Python, FastAPI, Node.js, React, Next.js, AWS