Ai Intern

About the Role
STAAD's AI clinical co-pilot currently generates session insights from live transcripts. The next step is grounding those insights in each therapist's own uploaded study material and past session history using a Retrieval-Augmented Generation (RAG) pipeline. As an AI Intern you will help design, build, and deploy this RAG system — from embedding documents to serving grounded responses in production.
What You Will Do
Implement end-to-end RAG pipelines: document ingestion, chunking strategies, embedding, vector storage, and retrieval.
Integrate the RAG layer with our existing AI analysis route .
Evaluate retrieval quality using standard RAG metrics .
Deploy and maintain the RAG service — containerisation, API exposure, and environment configuration on a cloud provider.
Explore hybrid retrieval (dense + sparse / BM25) and re-ranking to improve clinical relevance.
Document your pipeline decisions, evaluation results, and deployment steps.
What We're Looking For
Required
Solid Python skills; comfortable working with REST APIs.
Conceptual understanding of LLMs, embeddings, and semantic search.
Familiarity with at least one vector database (Pinecone, Weaviate, Qdrant, Chroma, or similar).
Ability to read and navigate an existing TypeScript/Next.js codebase.
Nice to Have
Hands-on experience with LangChain, LlamaIndex, or a similar RAG framework.
Exposure to Docker / containerised deployments.
Prior project (academic or personal) involving RAG, semantic search, or knowledge retrieval.
What You Will Learn
Full RAG lifecycle from data to deployed endpoint.
Working with clinical/sensitive data constraints (chunking strategy for therapy notes, PII considerations).
Integrating AI pipelines into a live production Next.js application.
Evaluation-driven iteration: measuring and improving retrieval and generation quality.