RAG & Agentic LLM Pipeline
A modular RAG pipeline over YouTube transcripts, extended into semantic search, a ReAct agent and an MCP server, packaged as a tested Python module.
ECE student @ RUET
Md. Adib Hasan
1st
Build with AI, Hackdays @ RUET
1 of 5
Teams funded, UIHP Innovation Cohort 5
AI engineer building LLM systems, RAG & agents
Open to AI engineering internships
Hackathons
5National hackathons
Work that ships
LLM systems, retrieval and the web apps around them, built end to end and measured.
- shipped
- live
Business Model Canvas Builder
A free, account-free Business Model, Lean and Value Proposition Canvas builder that exports real, editable Word and PowerPoint files, in seven languages.
28Countries in search within two weeks, Canvas Builder
- active
RUET Computing Society Website
The official RUET Computing Society website, a Next.js 15 platform with a custom HUD-style motion system and a Supabase backend in progress.
Anatomy of the Flagship Project:
RAG & Agentic LLM Pipeline
Chunking
YouTube transcripts split into chunks sized for retrieval
Embeddings
Each chunk embedded through the OpenAI API into namespaced Pinecone indexes
Retrieval
The question embedded the same way; the nearest chunks come back as the only context
Evals
pytest checks each seam, from ingest to answer, so a retrieval change cannot regress unseen
Experience
Where I have shipped under real deadlines, and what came of it.
Lab
Experiments, demos and prototypes: smaller than a Project, with no Case Study.