Sowrabha Rajesh

Designed and deployed LLM-based applications using OpenAI APIs and LangChain, improving contextual response quality by 35% for enterprise-grade use cases. Built RAG pipelines with vector databases such as Pinecone and FAISS to enable fast retrieval across 10M+ document records. Implemented prompt optimization and fine-tuning strategies that reduced operational token costs by 20% while maintaining strong model performance. Developed scalable MLOps pipelines using AWS SageMaker, Docker, and Kubernetes, reducing deployment cycles from days to hours, and collaborated with cross-functional teams to integrate ML models with production systems and data workflows.

Designed and deployed LLM-based applications using OpenAI APIs and LangChain, improving contextual response quality by 35% for enterprise-grade use cases. Built RAG pipelines using vector databases (Pinecone, FAISS) to enable fast retrieval over large-scale document systems (10M+ records). Implemented prompt optimization and fine-tuning strategies, reducing token costs by 20% while maintaining strong performance. Developed scalable MLOps pipelines with AWS SageMaker, Docker, and Kubernetes, reducing deployment cycles from days to hours, and collaborated with cross-functional teams to integrate models with production systems and data workflows.

Developed and deployed machine learning models (classification, regression, clustering) using Python, scikit-learn, and XGBoost, improving prediction accuracy by 18%. Built and optimized data processing and ETL pipelines using Apache Spark and SQL, handling 5M+ records daily for enterprise analytics reporting. Created interactive Power BI dashboards to improve stakeholder decision-making efficiency by ~20–25%. Delivered NLP-based solutions (BERT embeddings, TF-IDF, sentiment models) for document classification and text analytics, and supported the integration of ML outputs into business applications and reporting systems.