At Tektalis, I developed an enterprise leave management application with Microsoft Entra ID authentication and role-based access control.
At CtrlS, I built an end-to-end ETL pipeline on Databricks using PySpark, SQL, and AWS S3 to consolidate retail data into a unified Lakehouse. I implemented Bronze, Silver, and Gold layers with Medallion Architecture.
In a predictive caching project, I designed a system using Redis and LSTM that achieved an 80% cache hit rate. I also fine-tuned DistilBERT for domain-specific sentiment analysis, reaching 89.9% accuracy with LoRA.
I built an AI video assistant that transcribes videos with Whisper and generates summaries with LangChain and Mistral. It also supports RAG-based question answering with HuggingFace embeddings and ChromaDB.
