Rohan Vasudev

Ai Software Engineer @NewDay

London, GB

Collaborated with 4 engineers in AGILE env to design RAG-based GENAI tool for Customer Service, Customer Care and Fraud Teams, deployed to production on AWS(CDK) with 2500+ weekly users, increasing call speed 85% and calls handled per hour by 5× - Made a customer inquiry classifier with AWS Bedrock (Claude), LangChain, and Pydantic, achieving 83% accuracy- Enhanced customer inquiry classifier by using AWS Titan embeddings in ML clustering for semantic similarity, applying LLM inference only to low-confidence edge cases, improving accuracy to 97%- Created an ML evaluation system (SKLearn, S3) to track accuracy, F1, precision/recall, and confusion matrices for continuous live-model monitoring, expanding GENAI tool evaluation coverage by 33%- Engineered a Pydantic Schema layer to multi-environment AWS infrastructure - validating configurations, enforcing account and knowledge base rules – cutting configuration errors by 35%- Built a spaCy/SKLearn RAG pipeline to extract, chunk and embed documents, enabling semantic search with cosine similarity

UC San Diego

Bachelor of Science - BS, Mathematics and Computer Science