AI/ML/DL/LLM

AI/ML/DL/LLM is a deep-learning portfolio: a collection of notebooks and small services that demonstrate the canonical NLP, RAG, and computer-vision patterns, with a Gradio demo for each one so the model behaviour is browsable without standing up infrastructure.

The pieces: a YOLOv8 object detector on a custom small dataset; a sentence-transformer embedding pipeline persisted in ChromaDB; a RAG service that takes a question, retrieves the top-k chunks, and prompts an open-weight LLM; a Hugging Face fine-tuning notebook for a domain-specific text classifier; and a Gradio app that wires all four together so a single endpoint can be poked at end to end.

Each notebook is self-contained and runs on a single GPU. The Gradio demo ships in the repo and can be launched with one command, so a reviewer can verify the pipeline locally without the repo owner.

This entry does not link to a public repository. The notebooks and the Gradio demo were collected across a series of private projects; the entry here is the catalogue of patterns, not a single shipped app.