Personal Projects

Independent AI, deep learning and applied research work

Eleven independent projects covering deep learning, generative AI, natural language processing, computer vision, time-series forecasting and data visualization.

This is exploratory and research work rather than client delivery: architectures tested against public datasets, benchmarking studies, and applied experiments that later informed production engagements.

Representative work includes FoodPass AI, a retrieval-augmented chatbot answering real-time startup investment questions with LangSmith tracing; a hybrid residual LSTM for neural time-series forecasting; a LangServe infrastructure and LLM benchmarking study; a local PDF-RAG pipeline for offline knowledge extraction; an LLM resource optimization study; an autonomous financial multi-agent system; and a CNN-LSTM image captioning model.

Each entry documents the problem, the methodology, implementation notes and the results observed, with source repositories linked where the code is public.

Project index

• FoodPass AI: RAG Chatbot for Real-Time Startup Investment Due Diligence
• Neural Time-Series Forecasting: Hybrid Residual LSTM
• Cognitive Document Retrieval & GenAI Pipeline
• LangServe Infrastructure & LLM Benchmarking
• Localized Knowledge Extraction: PDF-RAG
• LLM Resource Optimization Study
• Global Trade Analysis Dashboard
• DC Urban Safety & Pattern Discovery
• Multi-Modal Financial Forecasting
• Autonomous Financial Multi-Agent System
• Visual Cognition: CNN-LSTM Image Captioning