Rasagna Kusuma

Developed enterprise analytics solutions using Power BI and SQL, reducing leadership reporting cycles by 35%. Designed ETL/ELT workflows to integrate, validate, and transform datasets into analytics-ready dimensional models. Implemented metadata mapping, schema validation, and data quality checks with 99.9% data integrity to support FDA 21 CFR Part 11 compliance. Delivered analytics training to 80+ users, increasing self-service reporting adoption by 45%.

Built LLM-powered applications and agentic AI orchestration workflows integrating LLMs, APIs, and retrieval systems for enterprise automation and intelligent data discovery. Developed Python microservices and REST APIs with FastAPI, and created AI-assisted pipelines for metadata enrichment, schema understanding, document processing, and data transformation. Implemented RAG/semantic search using embeddings and vector databases, deployed scalable services via Docker and Kubernetes on cloud platforms, and improved production reliability through model evaluation, monitoring, CI/CD, and validation workflows. Also trained ML models with Scikit-learn, XGBoost, and LightGBM, improving accuracy by 30%.

Developed Python-based ML and NLP pipelines on Databricks, Hadoop, and Hive for large-scale customer feedback analysis. Built distributed data transformation workflows using Python, SQL, and Spark. Implemented embedding-based semantic search and recommendation solutions using Word2Vec/GloVe with Pinecone and ChromaDB, improving precision by 15%. Integrated LangChain and LlamaIndex into retrieval workflows and exposed models via REST APIs (Flask/FastAPI). Deployed inference on AWS Lambda and GCP Vertex AI and supported preprocessing, classification, sentiment analysis, validation, and ongoing retraining/monitoring.