Aakash Sabharwal Aakash is an engineering leader in Enterprise Verticals at OpenAI. Previously, he was VP of Engineering at Scale AI, co-founded Lasso Labs, and led machine-learning and data-platform teams at Etsy. His work spans AI training infrastructure, model evaluations, search and ranking, and real-time ML systems.
Engineering Leader, Enterprise Verticals
VP of Engineering
Co-led a 150+ person generative AI organization across engineering, product, design, and ML research. Built human-data, coding-data, evaluation, and RL platforms; expanded frontier-model evaluations and led through the 2025 Meta transaction.
Co-Founder
Led foundational ML and data-platform strategy and teams of 15+ engineers and two managers. Built feature-data and streaming platforms, nearest-neighbor/vector-database services, and online experimentation systems.
Led context-specific ranking for search and advertising, from feature extraction and model training to deployment and inference. Built an AI system for identifying item style; the work was presented at KDD 2019.
Joined through the acquisition of Blackbird Technologies. Worked on search, AI, large-scale distributed systems, and systems infrastructure.
Engineer
Built search and ranking infrastructure, diversity ranking, Spark ETL, ranking models, and A/B testing systems using Solr, RabbitMQ, and Scala. Blackbird Technologies was acquired by Etsy.
Data Science
Built topic-modeling tools, ML insights ranking, and social-to-business KPI correlation metrics. Worked on terabyte-scale Hadoop ETL, real-time analytics, low-latency data slicing, and the Unfiltered product.
Worked with Professor Jaime Carbonell on unsupervised machine-learning problems in recruiting, including matching candidates with roles.
Teaching assistant for Introduction to Computer Systems.
Software Engineering Intern
Improved Hadoop data ingestion by 2–3× with OpenCSV. Built an Amazon EMR emulator for in-memory data summaries and improved diagnostic reporting for ingest jobs.
Software Engineering Intern
Designed and implemented a real-time video-trending pipeline using Scala and Storm. The system computed top videos using five algorithms and fed results into five website front ends. Built a staging environment using Hadoop jobs to replay logs and extract information.
