Shubhojeet Sarkar
Product Manager @ Meta
Menlo Park, California, United States
Actions
I am a Product Manager working with large-scale AI/ML systems, focused on how people discover, evaluate, and engage with content. My work sits at the intersection of product strategy, machine learning, and user experience—particularly around relevance, personalization, and long-term ecosystem health. I spend most of my time thinking about how AI systems understand user intent, how content is represented, and how product decisions compound at scale.
Area of Expertise
Topics
Evolution of the AI PM: From managing Products to managing Ecosystems
The modern landscape of consumer technology is defined by massive scale, intense competition for user attention, and the rapid deployment of artificial intelligence. On scaled platforms, Product Management for AI/ML features has evolved beyond incremental ranking gains to pioneering state-of-the-art techniques to improve personalization in turn building engaging ecosystems.
I intend to break down how successful AI/ML Product Managers translate complex user needs into high-impact product solutions, and build leverage through sophisticated cutting-edge models.
Leveraging LLMs to connect People to Answers, Not Just Links
For billions of people, digital communities hold a goldmine of authentic advice—from car repairs to gardening—but finding it often feels like searching for a needle in a haystack. This presentation details how we are using Large Language Models (LLMs) to eliminate the friction between users and the expertise they seek. We address two core frustrations: the failure of traditional search to understand natural language intent (like missing "cupcakes" when searching for "small cakes") and the fatigue of scrolling through endless comments to find a consensus. By blending "concept matching" with keyword search and deploying AI to summarize complex threads into instant insights, we help users find answers, not just links. We validate this improved experience using a novel "LLM-as-a-Judge" framework to ensure results are genuinely helpful, transforming platforms into intelligent engines of collective wisdom.
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top