Sanjeet Hajarnis

1. Developed a comprehensive Responsible AI framework, ensuring compliance with existing laws (NYC Law, EU AI Act, and Colorado Act) and anticipating future regulatory requirements 2. Published articles and whitepapers, establishing Eightfold's thought leadership in Responsible AI 3. Created a next-generation CoPilot that brought Eightfold's product to users within their preferred workflows, leveraging Generative AI agents. 4. Architected a framework-first approach for Large Language Models (LLMs), enabling rapid integration with both private (OpenAI, Gemini, Titan) and open source models (LLama, Mistral) - Developed an ensemble model that outperformed individual models - 3x faster development on CoPilot, 45% accuracy improvement in core matching, 100% accuracy improvement on resume parsing - Reducing costs by 30% 5. Led cross-functional collaboration between Engineering, Legal, and Marketing teams to position Eightfold as a thought leader in Responsible AI, directly contributing to the company's strategic objectives and market differentiation.

1. Drove AI improvements resulting in, $156M year-over-year savings across Eightfold's customer base, 2x reduction in time-to-hire, reduction in customer churn rate due to improved AI-driven matching accuracy, resulting in revenue retention. 2. Served as a subject matter expert on Eightfold AI on customers’ and prospects’ calls with Data Science, Legal, and Info Sec teams, helping to build trust and close deals faster. 3. Acted as a trusted advisor to legal teams, developing and negotiating AI-related SOWs that ensured compliance and clear guidelines. 4. Contributed to a 25% reduction in sales cycle time through timely technical support, objection handling, and cross-functional collaboration as a subject matter expert on AI. 5. Built a team structure based on focused high-performing pods, optimizing for technical depth and autonomy, and growing the AI team (4 to 30) 6. Mentored and coached team members to develop technical expertise and leadership skills, resulting in a 25% increase in internal promotions and a strong pipeline of future leaders.

1. Built all of Eightfold's AI from scratch ground up. 2. Scaled Eightfold's AI from initial POC and Pilot customers to large customers across various sectors, regions 3. Scaled Eightfold's AI to work across most professionally spoken languages 4. Pre-trained, Fine-tuned and Deployed BERT

1. Leader in the Marketplace Org, overseeing technical direction, innovation and strategy for Dynamic Pricing and Dispatch 2. Applied expertise in AI, Operations Research, and Economics to drive innovation and business impact in Uber's matching technology 3. Developed a forward looking dynamic pricing algorithm using Sparse CNN, Markov Chains, classification and regression models, resulting in number of trips improvement across 400 cities 4. Unified UberX and UberPool matching algorithms under a single objective function, affecting reduction in cost per incremental trip (CPIT)

1. Created initial version of FBLearner Flow, a pioneering machine learning platform that streamlined model development and deployment, democratizing access to AI capabilities across Facebook. 2. Contributed to FAIR's FAISS (Facebook AI Similarity Search) library, advancing the field of similarity search and indexing

1. Contributed to innovative projects, including: - Paper app: a novel, visually-driven news consumption experience - Hashtag launch: successfully rolled out hashtags across all Facebook, enhancing user engagement and discovery - Trends launch: launched hashtag and topic trends end to end

1. As an early engineer on the News Feed Ranking team, built the feed ranking system from scratch, evolving it from simple algorithms (Logistic Regression, Boosted Decision Trees) to advanced models (Factorization Machines, Deep Neural Networks, Computer Vision features on media content) 2. Designed and developed scalable infrastructure to support Facebook's massive scale of posts and requests, ensuring seamless user experience

- In browser Image Cropper using GWT and Javascript that allows advertisers to upload images and crop them appropriately for Display Ads on the Google Ad Network. - Added naive crop recommendations based on the image content

Building J2ME Test Applications to enhance the automated testing of the Blackberry device. Building a tool that would isolate the changes made on the successive blackberry softwares and analyze the data thus received to suggest filtered tests to run on the update

Worked on EBIF reader's reference implementation for set-top boxes. The project was done in J2ME