Senior Engineer at Fractal Analytics

Led RAG pipelines with LangChain using vector embeddings and smart document chunking (FAISS/Pinecone), improving LLM response relevance and retrieval accuracy by 30%. Developed and fine-tuned transformer-based NLP models using PyTorch and Hugging Face Transformers, applying improved prompt design and optimization techniques to increase prediction accuracy by 28% in production systems. Built high-performing ML models using XGBoost and TensorFlow, leveraging feature engineering and time-series forecasting methods to boost model precision by 15%. Implemented automated training and deployment pipelines with MLflow and Kubernetes, enabling scalable ML operations and reducing processing time by 40% while cutting cloud computing costs by 25%. Architected reliable ETL data pipelines to process structured and unstructured data using vector embeddings and scalable orchestration, achieving 99% reliability for downstream analytics workflows. Collaborated with stakeholders to deliver enterprise AI

Software Development Engineer Intern at Amazon

Trained, deployed, and maintained 5+ ML models for demand forecasting and anomaly detection using CNN and RNN architectures, improving prediction accuracy by 22% and enabling scalable inference for 50K+ transactions daily in AWS SageMaker. Built scalable data preprocessing and feature engineering pipelines with Python, Spark, and Bash, reducing model training time by 35% and refined hyperparameter tuning with Bayesian methods, increasing F1-scores by up to 18%. Implemented automated scripts to monitor model drift and integrated ML outputs using GitHub, Docker, AWS EC2, and CI/CD pipelines, reducing manual interventions by 40% and generating $500K+ projected annual savings.