At Infosys limited, I developed and deployed machine learning models for real-world use cases, improving prediction accuracy by 20% and reducing inference latency by 35%. I also built ML APIs with FastAPI and deployed inference services on AWS.
I designed RAG pipelines with LangChain and Pinecone, including document embedding, vector indexing, and retrieval. In a customer support assistant project, I processed more than 10K enterprise documents and helped reduce support ticket volume by 40%.
I evaluated the assistant with RAGAS, achieving 0.91 faithfulness, 0.88 answer relevancy, and 0.89 retrieval precision@5. I also used MLflow and CI/CD automation to support reproducible experiments and faster development cycles.
For my M.Tech project, I developed an Industry 4.0 digital twin for a rotordynamic fault simulator, integrating IoT sensors with Unity 3D. I deployed a TensorFlow 1D-CNN fault diagnosis model that achieved 96.2% accuracy and 95.8% F1-score.
