AI/ML Engineer

Apr 2021 - Jun 2023 | Hyderabad, India

• Delivered XGBoost credit risk classifier at 91% accuracy, reducing false positives 28% on 800K+ records; full model lifecycle managed in MLflow with bi-weekly automated retraining via CI/CD
• Built LSTM fraud detection pipeline on Kafka streams processing 50K+ transactions/minute at 94% precision, cutting fraud losses by 22% across financial portfolios
• Created Hugging Face + OpenAI NLP pipelines for financial report summarization — 35% reduction in analyst effort across 10+ enterprise clients
• Engineered CNN/LSTM deep learning models (TensorFlow, Keras, PyTorch) achieving 95% accuracy in image recognition and sequence prediction; containerized on Docker reducing latency 40% and infra costs 22%
• Optimized PostgreSQL and DynamoDB queries cutting pipeline runtime 50% on 5M+ record datasets; integrated 10+ systems into centralized ML feature store on AWS S3/RDS