At REVA University, I developed DeepCandle, a CNN-based system for real-time candlestick pattern detection that achieved 92% precision and was validated through historical-market backtesting.
For NEXUS Intelligence, I developed a machine-learning decision platform for SME financial data. It achieved 88% anomaly-detection accuracy and reduced manual forecasting time by 40% using Prophet.
I also co-authored research on an Edge AI system for legal assistance, and developed a RAG-based legal query assistant using LangChain, FAISS, and BERT embeddings. My work spans machine learning, NLP, computer vision, and time-series analysis.
