Mevin Jose - AI/ML Developer & Software Engineer

I build machine learning systems and full-stack platforms that hold up in production, covering agentic AI pipelines, real-time computer vision, event-driven microservices, and high-throughput data engineering workflows.

Featured Engineering Systems & Benchmarks

Key production benchmarks include 73.40% result equivalence and 100% SQL execution success on a 500-query Text-to-SQL benchmark with AST verification, a leakage-safe LightGBM forecasting pipeline trained on 58.3M historical records across 3,049 SKUs, an event-driven Kafka and FastAPI decision platform achieving 553 requests per second with 57% throughput gains, Mandate deterministic financial authorization control plane blocking 100% of unauthorized agent operations across 1,144 hostile scenarios, and Kavach AI serverless multi-agent dispute forensics on AWS Bedrock.

Experience & Education

AI Engineering Intern at Wadhwani Foundation, Web Developer Intern at Superhhero Learning, and B.Tech in Computer Science (AI & Machine Learning) at PES University, Bangalore.