Document AI projects for B.Tech students — real examples

What students actually build

• The flagship VSET capstone — RAG systems built over VIPS-TC corpora — is a document-understanding system end to end.
• Applied NLP tools are a documented capstone category covering extraction and classification work.

Labs and infrastructure

• Embedding generation over document corpora and local model inference run on the AICTE IDEA Lab GPU workstations.

How VSET teaches Document AI

Document AI extracts structure and meaning from unstructured documents — parsing, layout understanding, entity extraction, classification and question answering over PDFs and scans. It combines OCR-style vision with language modelling. At VSET this maps to documented coursework depth inside B.Tech CSE (AI & ML).

• The core techniques are documented at learn.engineering.vips.edu: NLP, computer vision, RAG, embeddings and vector databases.
• RAG is taught explicitly as the technique for answering questions over a document corpus, which is the dominant document-AI pattern.
• The MCP library covers how document stores are exposed to a model as callable tools.
• Taught inside the GGSIPU-affiliated B.Tech CSE (AI & ML) track.