May 2019 – Feb 2023 · 3 yrs 9 mos · San Francisco Bay Area

• I initiated Snap’s AI training platform effort initially as its founding engineer, and ultimately growing a cross functional team of ~20 Engineers (ML, Data, UX), UX Designers, and Project Managers.
• During my tenure the platform grew to support:
• Nearly all ML use-cases in the company, supporting 200+ ML engineers across 15 teams, and powering 30+ distinct ML modeling use-cases (including critical applications like Ads recommendation, organic content recommendation, search etc.).
• Between 1-2K training jobs per day on top of Petabyte level training datasets, including time-sensitive production model update pipelines responsible for $XB in revenue, and $XXXM in incremental revenue resulting from algorithmic improvements.
• Multiple interfaces including UX, Notebooks and python API/Command-line.
• Apart from managing the platform team, my responsibilities also included:
• Working with Snap’s finance team to establish and manage quarterly AI budgets (order of $XXM). And with customer teams to allocate it based on project priorities & set up enforcement mechanisms.
• Serving as the primary AI Training Platform POC for cloud providers like GCP and AWS.
• Formulating and communicating the platform strategy with the executive team.
• Evaluation of external AI/MLOps software providers, including running short term pilots.
• Interviewed ~250 prospective candidates to help grow AI Platform and other ML teams.

Nov 2018 – May 2019 · 6 mos · San Francisco Bay Area

• I was one of the founding Tech Leads (~4) tasked with designing the Dynamic Product Ads recommendation system for Snap. I worked with 30+ engineers and PMs across the company to successfully design, implement and roll out the first version of the system.
• Core technology was developed using product image embeddings, nearest neighbor vector databases, and recommendation ML models.
• Over the next 2 years the platform grew to deliver $XXXM in yearly revenues.

Nov 2017 – Nov 2018 · 1 yr · San Francisco Bay Area

• Co-founded a startup at the intersection of AI and Fashion e-commerce.
• We developed AI (language model and vision) based 'complete the look' technology to automatically curate an outfit by grouping aesthetically compatible fashion products from a retailer's website.
• Expert evaluation by human stylists found 83% of generated outfits were good quality.
• Built a small team of UX engineers and fashion stylists.
• Initiated pilot trials at a mid sized fashion retailer.
• Acqui-hired by SNAP Inc. to help power their product Ads recommendations.

Oct 2017 – Present · 8 yrs 6 mos

• I do limited consulting to help both startups and large companies develop their AI initiatives.

May 2016 – Oct 2017 · 1 yr 5 mos · Sunnyvale

• Lead and managed the machine learning based search relevance and ranking efforts at WalmartLabs impacting $XX billion in revenue.
• Directly contributed revenue lifts of over $XXX million.
• Typical projects areas were: Learning to rank, Text retrieval models, Multi-stage ranking, Search result Personalization, Machine learning experimentation framework design, Multi-armed bandits (explore-exploit), Large scale mining of query-product user engagement graphs, Statistical query interpretation, Crowdsourcing, Designing online and offline evaluation metrics for search

Oct 2013 – May 2016 · 2 yrs 7 mos · Sunnyvale

• Lead multiple ranking projects related to learning to rank, top-k retrieval, offline & crowd-sourced search evaluation, statistical query interpretation, incorporating user engagement signals etc.. Responsibilities included leading teams of 1 − 2 engineers, and owning projects end to end; from conception, to deployment, to production.

May 2011 – Aug 2011 · 3 mos · Hawthorne NY (Healthcare Transformation Group)

May 2009 – Aug 2009 · 3 mos · Microsoft Research - Redmond (Text Mining Search and Navigation Group)

Jan 2008 – Jan 2013 · 5 yrs