HRBuddy
AI / RAGHR knowledge-retrieval application that chunks a 64-page policy document into approximately 150 semantic chunks and returns grounded answers to natural language queries.
PROJECT ARCHIVE
A working record of AI applications, local-first tools, and experiments across document intelligence, agents, and retrieval.
HR knowledge-retrieval application that chunks a 64-page policy document into approximately 150 semantic chunks and returns grounded answers to natural language queries.
Private document intelligence platform running Gemma 4B on-device for secure, cloud-free compute, embeddings, and prompt execution without cloud dependencies.
Autonomous terminal AI agent built with Python and LangGraph that auto-executes safe commands and uses configurable guardrails for higher-risk operations.
AI-powered resume analytics application using Streamlit and a carefully prompted Azure OpenAI model, deployed on Azure App Service.
Automated invoice digitization with NLP and OCR. Took a RAG pipeline from prototype to production using ChromaDB, hybrid retrieval, and cross-encoder reranking. Refactored invoice parsing into an agentic workflow with tool calling, model evaluation, and prompt engineering.
Applied computer vision and audio classification models with image processing pipelines for industrial inspection and defect detection.
Processed unstructured data pipelines with Pandas and implemented computer vision and text extraction workflows with OpenCV and Pytesseract-OCR.
Generative AI, LLMs, RAG, Agentic AI, Tool Calling, Prompt Engineering, Transformer Models, ChromaDB, NLP, OCR, Computer Vision.
Python, LangGraph, LangChain, PyTorch, TensorFlow, Keras, OpenCV, Pytesseract-OCR, Pandas, NumPy, Streamlit.
Microsoft Azure, Azure OpenAI Service, Azure App Service, Docker.