Applied AI & Machine Learning Solutions
Moderate to High
~40 Hours Build
Contextual Document Intelligence Assistant with RAG & Vector Search
Build a production AI web assistant that ingests company PDF manuals, chunks and embeds text into a vector database, and generates factually-grounded answers via LLM prompt chaining.
Required Deliverables
What You Must Build to Graduate
- ✓ Document ingestion pipeline with text extraction and semantic chunking.
- ✓ Vector database setup (Chroma / Pinecone / pgvector) with cosine similarity retrieval.
- ✓ Prompt chain mitigating hallucinations and citing source page numbers.
- ✓ Clean web interface allowing users to upload documents and query answers.
Verification Rubric
Evaluation & Quality Criteria
AI Engineering & Retrieval (40%): High semantic accuracy and prompt grounding.
Architecture & Code (35%): Efficient token usage and clean Python/TypeScript API.
User Interface & Speed (25%): Streaming responses with clear source citations.
Proof of Work Submission
Log in or create a learner account to submit your project proof of work for review and verified badges.
Log In to Submit