Avinaash Anand

About Avinaash Anand

AI Product Manager building evaluation and safety infrastructure for GenAI products at Uber scale (XX M+ monthly customer contacts). I specialize in LLM-as-Judge evaluation systems, AI guardrails, synthetic data generation and Conversational Simulation, critical infrastructure that enables safe, fast AI deployments.WHAT I\'VE BUILT:At Uber (Current): → Shipped LLM-as-Judge evaluation platform with XX reusable metrics, reducing feedback loops from weeks to minutes → Built conversation simulation engine cutting evaluation dataset creation time to hours, enabling AI agent expansion to across global markets → Launched AI guardrails (PII detection, jailbreak prevention, high-risk topics, communication guidelines adherence, content safety) across GenAI products → Awarded \"See the Forest and Trees\" — 1 of 2 IC awards at Org All HandsAs Founder (Fluffy): → Built RAG-based code assistant scaled to 100+ active users and 2 paying design partners → Engineered 2x improvement in retrieval relevance using custom rerankers and LLM-as-Judge evaluationResearch & Competitions: → 2nd place, NeurIPS 2024 Concordia Contest (Multiagent Systems) — co-author on benchmark paper → 9th place, NeurIPS 2024 CLAS for LLM Safety ChallengeMY APPROACH:I bridge deep technical understanding of LLMs with product craft. I\'ve shipped code myself when timelines were at risk, designed LLM-as-Jury architectures with multiple reasoning models, and authored red teaming strategy. I believe the best AI PMs understand both the model and the user.TECHNICAL DEPTH:LLM Evaluation | Prompt Engineering | Synthetic Data Generation | AI Guardrails | Red Teaming | RAG Systems | Transformers | PyTorchOpen to connecting with AI/ML practitioners, discussing GenAI product challenges.

Experience

1. Product Manager

   Uber

   Apr 2025 — Present

   Leading AI evaluation and safety infrastructure for Uber\'s GenAI Support products supporting XX M+ monthly customer contacts. Incubated 0→1 platforms that enable faster iteration cycles and safer deployments across all Customer Support AI agents.KEY ACHIEVEMENTS: → Built LLM-as-Judge evaluation platform with xx+ reusable metrics; reduced feedback loops from weeks to minutes at 90%+ accuracy → Shipped conversation simulation engine cutting evaluation dataset creation time to hours to enable global expansion across markets → Launched AI guardrails for (PII detection, jailbreak prevention, high-risk topics, content safety and communication guideline adherance) establishing safety layer across all GenAI products; led cross-org alignment with EngSec, Platform, Legal & Ops → Authored Red Teaming Strategy and drove company-wide platform adoption collaborating with EngSec, Operations and Legal counterparts. → Awarded \"See the Forest and Trees\" — 1 of 2 IC awards at org All HandsSCOPE: LLM Evaluation, AI Guardrails, Synthetic Data Generation, Red Teaming, Cross-functional Leadership

Education

• Indian Institute of Technology, Bombay

  Bachelor of Technology (BTech), Chemical Engineering

  2013 — 2017

• Indian Institute of Management Ahmedabad

  Master of Business Administration - MBA

  2017 — 2019

• Maharishi International Residential School

  High School Diploma

  2011 — 2013

Skills

• Photoshop
• Html
• Microsoft Office
• Research
• Powerpoint
• Microsoft Excel
• C
• C++
• Matlab
• Microsoft Word