Customer Support Copilot
Enterprise assistant that answers support queries with RAG, LangGraph agent workflows and persistent conversation memory.
RAGLangGraphLangChainVector DB
Hi, I'm Pankaj Kumar.
I build RAG applications
AI/ML Engineer Intern at TheSmartBridge, working on Generative AI, retrieval-augmented generation and multi-agent systems. My base is data science and analytics, so I care about evaluation, data quality and shipping things that work.
Open to AI/ML roles · Delhi NCR20+AI/ML projects on GitHub
34Licenses & certifications
#1Branch topper, B.Tech CSE (DS)
3Talks & conference presentations
What I work on
Four areas, with GenAI as the main focus.
Generative AI & RAG
Document ingestion, chunking, embeddings, retrieval and grounded generation with source citations.
LLMsLangChainChromaDBFAISS
Agentic AI & MCP
Multi-agent pipelines with planning, tool use, memory, and MCP connectors to external tools and data.
LangGraphAgentsMCPTool use
Computer vision & multimodal
Detection, tracking and text-plus-image models for safety, agriculture and misinformation.
YOLOv8OpenCVPyTorch
Data science & analytics
EDA, feature engineering, forecasting and dashboards that turn data into decisions.
SQLPandasPower BITableau
Featured projects
ResearchMind AI
Multi-agent research system: a LangGraph pipeline plans questions, searches the web, analyses findings against your own PDFs, and writes a report.
LangGraphAgentsRAGPython
QuizGen AI
Turns PDF/DOCX/TXT notes into grounded, difficulty-controlled MCQ quizzes with a full RAG pipeline (parse, chunk, embed, retrieve, generate).
RAGChromaDBFastAPIStreamlit
How I build AI systems
Interactive workflow snapshots — select a workflow to explore its stages.
01IngestPDFs, text, data
→
02Chunk & embedSemantic vectors
→
03RetrieveVector search
→
04GenerateGrounded answer
A retrieval-augmented generation flow that grounds responses in source documents.