Anish Reddy Yellakonda

Data Scientist @Wells Fargo

NY, US

Engineered and deployed machine learning models using XGBoost and LightGBM to classify high-risk financial transactions based on login velocity, IP mismatch, and transaction context. Improved detection rate from 9/10 to 11/12 in test batches, reinforcing compliance trust.• Created sector-specific visualizations using NumPy, Matplotlib, and Seaborn across 6 fiscal quarters to uncover irregular market behaviors. Insights enabled timely asset rebalancing within 2 weeks, enhancing portfolio agility during volatile periods.• Analyzed behavioral patterns from over 12 million transactions using SQL and Python, improving fraud-tagging precision by 18% and strengthening monitoring protocols for high-risk transaction clusters.• Developed a deep learning LSTM model using PyTorch to detect sequential anomalies in financial activity logs, reducing false negatives by 21% and exposing previously undetected fraudulent behavior.• Fine-tuned a BERT-based NLP model to auto-classify and route customer complaints from emails and call transcripts, increasing classification accuracy from 70% to 85% on unseen cases and expediting fraud risk escalation.• Revamped multi-stage data pipelines using Python, REST APIs, and Apache Airflow to automate validation and integration. Reduced data latency from 4 hours to 1.5 hours daily, providing fresher insights for analytics teams.• Built a LangChain-powered GenAI explainer to interpret fraud model decisions, reducing stakeholder prep time by 60% and improving transparency during audit and model governance reviews.

University at Albany

Master's degree