Devanshu Brahmbhatt
AI Product Manager at Skyflow | PaperTalk.io | Ex- Rakuten
San Francisco, California
Overview
Work Experience
Product Manager - AI, LLM Privacy
2024 - Current
■ Leading an AI service to extract and redact sensitive data from unstructured formats (images, audio, text), processing over 1TB of customer data, continuously enhancing APIs for a better developer experience, and integrating with a vault for secure entity storage. ■ Enhanced infrastructure capabilities by integrating advanced ETL support, reducing overall processing time by approximately 53% and achieving more than double the previous speed. ■ Steer product vision & strategy hand-in-hand with Solutions, Sales, and GTM teams, shaping market positioning, defining privacy-first AI features, and leading thought-leadership through blogs, POCs, presentations, and competitive evaluations
Product Management Intern
2024 - 2024
■ Researched LLM tokenizers for prompt evaluations, influencing key product design decisions, particularly in selecting token formats that maximize LLM effectiveness while ensuring no personal information is sent to LLMs. ■ Led the development and pilot implementation of LLM guardrails for Skyflow's vault, successfully delivering enhanced data security to major customers, focusing on validating both input and output to protect sensitive information.
Skyflow provides a data privacy vault with an API interface.
Raised $97,000,000.00 from Canvas Ventures, Firestreak Ventures, Khosla Ventures, Foundation Capital, Mouro Capital, MS&AD Ventures, Gokul Rajaram, Canvas Ventures, Foundation Capital and Insight Partners.
Co-Founder
2023 - 2024
■ Launched PaperTalk.io to help students, researchers, and professionals easily find, understand, and apply research in real-world applications, enhancing accessibility with AI explanations. ■ Led hiring for a developer and UI/UX engineer, working with two co-founders. Managed a team of 4-person, enhancing functionality and user experience, increasing engagement by 78%, and scaling to over 5,000 users.
Research Intern
2023 - 2024
■ We designed 'QubiCSV: An Open-Source Data Management Platform' This platform helps scientists store extensive quantum computing data and allows visualization of results for improved qubit performance. We are among the earliest efforts in the quantum field to implement data versioning at the database level, which makes this platform unique. ■ 3000+ article accesses, Altmetric score 24 (top 9 % of articles of similar age), featured in 3 news outlets and 10+ social posts. ■ Utilized DoltDB for version control, enabling scientists to maintain individual branches for easier collaboration and reducing manual file maintenance by 50%. The platform includes a dashboard and a Jupyter Notebook for access.
Berkeley Lab is a national laboratory that creates advanced new tools for scientific discovery.
Software Engineer Intern
2023 - 2023
■ Led the design of an MLOps Project utilizing LangSmith & RAGAS to evaluate AI agent performance. ■ Implemented RAGAS & G-val LLM evaluation metrics on the existing system to enhance performance assessments.
Graduate Research Assistant
2022 - 2023
■ Led the design and creation of a data-storage pipeline for a QubiC, initiating the project in collaboration with the LBL lab. ■ Processed EDF data, converted it to CSV, effectively labeled seizure data, and achieved 97% accuracy in identifying seizures Thesis Advisor: Dr. VP Nguyen
Software Engineer I
2021 - 2022
■ Worked with Rakuten's Energy Team on Denki and Rakuten GAS products in Japan, overcoming language barriers by understanding legacy code and system through Japanese documentation, effectively collaborating, and delivering multiple features within agreed timelines. ■ Developed modules for Japan's Rakuten Gas system using Spring Boot and Laravel Framework, successfully implementing microservices architecture by converting the entire structure from monolith to microservices, while utilizing tools like Bitbucket, Docker, and Kubernetes for efficient code deployment in Dev, STG, and Production environments. ■ Architected Spring Boot API for test data generation, independently designing, coding, and testing the solution. Reduced engineering time by 10,000+ man-hour
Technical Intern
2020 - 2021
■ Took sole responsibility for designing, developing, and testing Spring Boot APIs for a product that generates test data for software testing purposes. ■ Created an iOS mobile application prototype for the Denki web application, a major product of Rakuten Japan in the low-voltage electricity supply service. ■ Organized and successfully completed feature developments, including functional and unit testing, utilizing the WBS (Work Breakdown Structure) technique.
SHOPVAULT is an online retailer of men's and women's fashion and accessories.
Lead - Developer Student Club DDU
2019 - 2021
■ Team Leadership: Led a 30-member team and orchestrated over 20 campus events and live seminars, providing 1500+ students exposure to Google Technologies, and fostering a development culture and community for college students. ■ Innovative Initiatives: Pioneered execution of key initiatives including hackathons and coding boot camps, driving a 25% increase in DSC membership.
Full Stack Engineer
2020 - 2020
■ Contributed to Tailbuds, a start-up focused on dog care and dog accessories, by working with technologies such as Node.js, Express.js, Angular, MySQL, and Ionic5. Developed CRUD API for seamless interaction with a complex database schema, following an Agile process and implementing CI/CD practices.