Sagar Jain

AI Researcher

Mountain View, California, United States13 yrs 10 mos experience

Key Highlights

Led advanced AI research at OpenAI.
Pioneered personalization systems at Google.
Expert in large-scale machine learning models.

Stackforce AI infers this person is a SaaS expert specializing in machine learning and personalization systems.

Contact

Skills

Core Skills

Machine LearningPersonalizationRecommender SystemsReal-time Systems

Other Skills

User ModelingData AnalyticsC++CC#F#HaskellJavaPythonLinuxLLVM

About

I am currently an AI Researcher @ OpenAI. Previously, I led multiple applied machine learning teams at Meta and Google to advance large-scale Language Models, Personalization and Recommender Systems research. I enjoy solving complex real-world problems. Aside from work, I am a weekend warrior who is fond of ultra distance trail running, experimenting with coffee and photography.

Experience

13 yrs 10 mos

Total Experience

4 yrs 4 mos

Average Tenure

8 mos

Current Experience

Openai

Member of Technical Staff

Sep 2025 – Present · 8 mos · San Francisco, California, United States · On-site

Frontier AI research

Instagram

Research Principal Engineer

Sep 2021 – May 2024 · 2 yrs 8 mos · Menlo Park, California, United States · On-site

Uber AI/ML Instagram Recommendations TL

Google

Research Senior Staff Software Engineer, Google Brain

Feb 2012 – Sep 2021 · 9 yrs 7 mos · Mountain View, California, United States

2017 - 2021:
Led a team of engineers to advance user sequence modeling and scale the Personalization research effort for improving user experience in Ads. The personalization infrastructure and models, I led and designed -- unified and integrated previous personalization infrastructure and have been applied to many ranking and conversion models within Ads. The end-to-end platform (from training to serving) enabled much faster user modeling research iterations and landings in Ads. Beyond scaling and user modeling improvements, I also worked on addressing the feedback loop effect between the user behavior learning and the system policy learning. My team has also delivered many recommender systems ML modeling improvements in Ads (ranking and conversion) and YouTube (retrieval and ranking) leading to significant impact for Google.
2014 - 2017:
Co-led, co-designed and built large-scale Recurrent Neural Recommender system for YouTube. Challenges include modeling user temporal activity information and incorporating other context and personalization features and scaling / serving in real-time. I also co-led and worked on solving several key problems of addressing clickbaits recommendations, whole-page optimization with diversifying recommendations.
2012 - 2014:
Led, designed and built the real-time and centralized Google+ user storage infrastructure for serving user signals at low latency to the online production systems (e.g. ML models) as well as for batch processing needs like data analytics pipelines. The infrastructure that I built stored both user (node-based) and edge-based data consistently in the same storage, and met massive scaling needs for both real-time and batch use-cases. Delivered a large impact for Google+ enabling many personalized ML applications (e.g. Feed Ranking, Friends Suggestion, Notifications) and data analytics dashboards.

Machine LearningUser ModelingPersonalizationRecommender SystemsReal-time Systems