Projects
A collection of applied work that demonstrates how AI, machine learning, and data readiness come together in practice. Each project highlights real-world challenges and solutions, emphasizing clarity, transparency, and measurable impact. Every build is supported with open code, detailed explanations, and visuals that make complex concepts accessible.
AI-Powered Document Search with Vector DBs
Description:
Built a semantic search system using embeddings and a vector database to retrieve the most relevant passages from unstructured documents. Demonstrates how LLMs + retrieval can replace brittle keyword search.
Tech Stack: Python · Pinecone · Hugging Face · Streamlit
Lightweight AI Q&A with Small Language Models
An experiment in running efficient, privacy-preserving language models locally using Ollama, paired with a vector database for fast semantic search. This project explores how Small Language Models (SLMs) can deliver practical results in scenarios where resources are limited or data must remain on-device—such as healthcare, finance, or edge deployments.
Tech Stack: Ollama, Python, FAISS / Pinecone, LangChain
Visual: Diagram showing local SLM + vector DB workflow
Link: Full Project Page
GPU Performance Comparison for AI Tasks
A straightforward project testing different NVIDIA GPUs on common AI workloads such as image classification and text generation. The focus is on measuring speed, efficiency, and cost to show how hardware choices impact AI development.
Tech Stack: Python, PyTorch, Hugging Face, CUDA
Visual: Simple bar chart (training time across GPUs)
Link: Full Project Page
