Adhya
Dagar
San Francisco, CA
Experienced Product Manager with a strong background in AI, Machine Learning, and product development, specializing in designing and evaluating generative AI models and agentic systems. Proven ability to drive user acquisition, retention, and product-market fit through data-driven strategies and cross-functional leadership. Seeking to leverage expertise in building scalable AI solutions and transforming research into reliable products.
Scroll to explore
Experience
My professional journey and the amazing teams I've been part of
Product Manager, Model Evaluations, Adobe Firefly
Adobe
Sep 2025 — PresentCurrent
Leading Firefly Enterprise Evaluations - Built human and automated evaluation systems for image, video, and audio generation models across video reframe, vectorization, image and video editing, and custom LoRAs models.
Partnered with creative professionals and enterprise customers to define production-quality standards, translating expectations into measurable rubrics determining model readiness and launch decisions.
Driving cultural transformation toward evaluation-driven product development – training PMs, standardizing eval study templates, launching enablement playbooks and embedded quality gates across Adobe’s generative AI workflows.
Scaling Agentic Evaluation & Improvement – Built evaluation frameworks and test harnesses to assess and improve agentic capabilities across multimodal workflows.
Product Manager (Research & Evaluations)
Pareto.AI
Mar 2024 — Aug 2025
Designed and deployed agent evaluation infrastructure creating programmatically verifiable RL environments across 15+ professional domains (e.g., Data Science, Healthcare Operations, Finance) and 10+ OSS tools (e.g., OpenEMR, SuiteCRM, Metabase) to evaluate multi-step agent workflows, tool use, and complex reasoning, leading 20-member team to establish new standards for real-world agent performance assessment.
Designed a comprehensive legal eval framework from proof of concept to deployment, alongside legal domain experts, strengthening frontier model legal reasoning across contract review, regulatory analysis, and case law.
Built an automated data quality engine with a comprehensive evaluation taxonomy including LLM judges (instruction following, complexity), statistical checks (preference distribution, semantic similarity), and fraud detection reducing manual QA from 100% to 20%.
Drove new frontier lab partnerships by researching emerging evaluation needs, scoping pilots, and designing 10+ tailored proposals that converted into long-term collaborations and revenue growth for Pareto.
Reduced manual QA from 100% to 20% through automated data quality engine.
Established new standards for real-world agent performance assessment with a 20-member team.
Converted 10+ tailored proposals into long-term collaborations and revenue growth.
?token=pk_cbQTqhuzT2OyDtYy16C0LA){kind=logo}
Product Manager
BitClass (Live cohort-based learning app)
Jan 2022 — May 2022
Drove 170% user acquisition growth from 78K to 210K monthly registrations through quick-checkout system that streamlined purchase flows, while launching short-form video content feature repurposed from recorded live sessions that boosted user retention by 60% across 1.3M user base.
Optimized user experience through data-driven redesign using A/B testing and conversion funnel analysis to improve course discovery, home screen navigation, and checkout flows, delivering measurable increases in user engagement and purchase conversion rates.
Drove 170% user acquisition growth.
Boosted user retention by 60% across 1.3M user base.
Delivered measurable increases in user engagement and purchase conversion rates.
Education
Where I built the foundation of my expertise
Carnegie Mellon University
Master of Science in Artificial Intelligence and Innovation
Class of 2024
Vellore Institute of Technology
Bachelor of Technology in Computer Science Engineering
Class of 2020
Links
No links provided