CDP Pipeline
High-scale Customer Data Platform on AWS Lambda, EventBridge, Apache Unomi and Elasticsearch — ingesting 150k+ events/sec.
// about
I build production AI systems — MCP servers, agentic workflows, LLM orchestration, and serverless event-driven backends on AWS and GCP. I like hard systems problems and shipping things that hold up in production.
At Vyg.ai I architected a high-scale CDP (Customer Data Platform) on AWS Lambda, EventBridge, Apache Unomi, and Elasticsearch handling 150k+ events/sec, and run an AI pipeline over 2,500+ conversations a day that cut inference cost in half. I've shipped internal and customer-facing MCP servers that let AI assistants query brand data straight from ChatGPT and Claude, and automated our dev and maintenance workflow with AI bots for monitoring and bug-fixing.
Sylhet, Bangladesh · remote across US + BD teams
// selected work
High-scale Customer Data Platform on AWS Lambda, EventBridge, Apache Unomi and Elasticsearch — ingesting 150k+ events/sec.
OAuth-protected MCP servers that let AI assistants query brand data directly from ChatGPT, Claude and other clients.
Production AI pipeline processing 2,500+ conversations/day with OpenAI tool calling and structured outputs.
LangGraph-powered meeting assistant — an agentic chatbot over the full meeting corpus.
Waste-management platform for Dhaka City Corp — RBAC, Google Maps, Socket.io live tracking, PDF reports.
Text-to-reel generator that turns a prompt into a finished short video.
Natural-language expense tracker — log spending by just typing it.
// experience
// stack
// proof
Shahjalal University of Science & TechnologyB.Sc. in Software Engineering
// let's talk
If you're hiring for hard AI-systems problems, I'd love to hear about it.
© 2026 Shawon Majid