I TURN MESSY DATA INTO SYSTEMS THAT MATTER.
Offline enterprise RAG with hybrid BM25 + vector retrieval, cross-encoder reranking, and zero external API exposure.
10-year probabilistic demand forecast using DeepAR with ensemble NWP weather covariates and a 4-model comparison dashboard.
NLP framework extracting thematic + sentiment signals from 68K S&P 500 news articles, validated against real sector returns using Alphalens.
Near-real-time monitoring pipeline that replaced hourly manual SQL checks with automated fault detection and alert dispatch.
AI-driven stock advisor aggregating financial news, fundamentals, and economic signals into real-time LLM-generated investment analysis.
Classification system predicting rating tiers for unrated Airbnb listings using PySpark on Databricks with XGBoost and Random Forest.