Technical Program Manager | AI/ML Quality & Evaluation | IoT Product Launches
9+ years driving cross-functional delivery of ML systems, GenAI pipelines, and IoT products at Amazon. I build frameworks that bring structure to ambiguous AI problems - from evaluation loops to launch readiness.
Tools: Python, RAGAS, AI/ML, GEN AI, GitHub Actions, AWS (SageMaker, Bedrock), Jira
Early-stage AI/ML startups often struggle with: "How do I know if my model is good enough?" This framework gives founders a structured way to define their quality bar, identify risks before they become blockers, and build evaluation loops without needing third-party annotators or large QA teams.
Python · LLM-as-Judge · Evaluation Frameworks
RAG systems can hallucinate, retrieve wrong context, or give irrelevant answers and most teams don't know until users complain. This project implements automated evaluation using RAGAS to catch quality issues before production with repeatable CI-integrated tests.
Python · RAGAS · LangChain · GitHub Actions
-
AI evaluation at scale : How do you know your AI is getting better, not worse, with every release? I'm building toward evaluation systems that scale without scaling human reviewers - synthetic test users, LLM-as-judge, and severity-calibrated testing to catch critical failures before users do.
-
Responsible AI compliance : The EU AI Act takes full effect August 2026. I'm exploring what that means practically for teams building AI products today — bias testing, transparency requirements, and building compliance into the dev process rather than bolting it on at launch.
-
Startup advisory : Helping early-stage AI/ML teams define what "good enough" looks like, identify risks early, and build a repeatable process for shipping confidently.
-
Panelist — Project Management World (PMW): "Cancelled Projects — What We Learned and How We Moved Forward" (April 2026)
- eranderashmi25@gmail.com
- 🎤 Available for speaking engagements | 📋 Open to full-time TPM opportunities
💬 Open to advising early-stage AI teams on evaluation strategy and launch readiness — reach out if that's you.