Back to portfolio

AI Integration
RAG, CAG & Fine-Tuning Research
- Role
- AI Integrator · Researcher
A college research project exploring retrieval-augmented generation (RAG), cache-augmented generation (CAG), and fine-tuning workflows. Used GPT-4o mini as the base model, Pinecone for vector retrieval, OpenAI APIs for RAG/CAG pipelines, OpenAI cloud fine-tuning for custom models, and Node.js to orchestrate the full experimentation stack.
Keynotes
- Implemented RAG pipelines with Pinecone vector storage and OpenAI embeddings.
- Explored CAG patterns to reduce latency and repeated retrieval cost.
- Fine-tuned GPT-4o mini via OpenAI cloud and compared outputs across approaches.
- Built a Node.js service layer to unify API calls, prompts, and evaluation flows.
Technologies
- Node.js
- TypeScript
- OpenAI API
- GPT-4o mini
- Pinecone
- RAG
- Fine-tuning