Varshine Sridharan

AI Product Lead- Deep Research Agent & Gen AI Experiences @ Microsoft

About

I'm a high performing AI product manager with deep, technical experience in building and shipping AI, ML products and Agents across enterprise and consumer ecosystem. My work spans agentic systems, model agility, evaluation frameworks, reasoning quality, responsible AI deployment, RAG systems and compete evaluations. I translate cutting-edge AI into tangible products to help users be productive and deliver outcomes at scale. I thrive in areas of high ambiguity, while leading cross-functional teams and partnering with smart builders to deliver impactful results in the intersection of AI, Business and Design. I have built 0-1 products and scaled usage & adoption of experiences and products to hundreds of millions of users, while driving retention & growth opportunities. I recently built 0->1 agents for productivity and launched the Researcher agent in Microsoft 365 Copilot helping millions of users and businesses synthesize knowledge effectively. Previously, I led product for Gen AI powered experiences for Microsoft Dynamics 365 Customer Service Copilot products where I drove product efforts to reimagine and build AI native Customer Service and Contact Center apps, saving hundreds of millions of dollars for Microsoft support business. Earlier, I built API & Data products at Twilio Segment to power adaptive, personalized customer experiences with good data. I enjoy vibe-coding apps, reading the latest research papers and brainstorming tech ideas which I actively share here: https://x.com/VarshineSri .Feel free to reach out if you're building the next wave in tech!

Country

City

United States

Industry

Computer Software

Skill

AI Agents, Multi-agent Systems, AI product evaluations, Python (Programming Language), Retrieval-Augmented Generation (RAG), Prompt Engineering, Large Language Models (LLM), Product Leadership, Mobile Product Development, Machine Learning, Artificial Intelligence for Business, Product Roadmap, Generative AI, Software Product Management, Software Development, Program Management, Artificial Intelligence (AI), CRM Integration, Predictive Modeling, Prediction

Experience

AI Product Lead- Deep Research Agent & Gen AI Experiences

Microsoft

2024-5 - Present · 2 yrs 5 mos

Redmond, Washington, United States

Product Lead for Microsoft Researcher Agent - a fine-tuned “deep research” assistant that is powered by OpenAI reasoning models (along with RAG solution for grounding) with iterative chain of thought reasoning to reason over enterprise & web data to create comprehensive research reports, competing against ChatGPT Deep Research and other deep research tools. -Built and shipped Microsoft Researcher agent 0→1 impacting hundreds of millions of M365 copilot users; Defined vision and user experience & shipped V1 on an accelerated timeline by partnering across Design, Eng, Research, UX, Marketing, Legal, and GTM. -Reasoning quality: Spearheaded the end-to-end quality evaluations for the agent responses, model behavior, web retrieval by combining LLM-as-judge and human-in-the-loop and defined evaluation set. Partnered with Applied Scientists to improve reasoning quality(iterative planning, search & retrieval) and reduce hallucinations rates and cost for enterprise workloads. -AI Safety: Led end-to-end AI Safety/RAI product reviews (with OpenAI) prior to external release, by partnering with Security, Privacy, Legal, AI Safety Scientists and AI Red Teamers and established safety evaluations effort for AI agents. -Competitive benchmarks: Built a compete eval strategy using open source benchmarks to score web & work response quality against OpenAI, Anthropic (Claude), Google DeepMind (Gemini), Perplexity, and xAI; translated insights into roadmap bets. -Drove product rituals and OKRs, presented to C-Suite & SLT to align across exec stakeholders, drove customer demos and feedback loops and delivered brownbag sessions on evals & AI quality.

Senior Product Manager- Dynamics 365 Copilot for Customer Experiences

Microsoft

2022-12 - 2024-4 · 1 yr 5 mos

Redmond, Washington, United States

LLMs for CRM App- Dynamics 365 Copilot- Copilot serving customers in collaboration with humans! -Drove ideation, user research and productization of conversational AI products in CRM by leveraging AI/ML models for support automation workflows, resulting in 30% reduction in case handling time and saving >$3M/yr in support costs. -Lead development, Productization & Growth of Gen AI based Copilot Product offering in Dynamics 365 Customer service for consumer support businesses. -Shipped v1 Dynamics Copilot for Service by partnering with Applied Science & Research, UX teams, marketing, customer support & operations to automate customer service scenarios by leveraging GPT models and LLMs to infuse AI into Dynamics 365 customer service app, increasing MAU & WAU by 55% & Helpful Response rate by 30%.