I build AI products that close the gap between what frontier models can do and what enterprises can reliably deploy. My work spans enterprise AI, privacy infrastructure, developer APIs, agent systems, and products for sensitive data.

Portfolio · LinkedIn · Google Scholar · X · Medium

• Leading 0→1 AI product work for Skyflow Detect, helping enterprises identify and protect sensitive data across AI, data, and security workflows.
• Building and operating agents to understand what production systems need beyond a prompt: context, tools, permissions, memory, evaluation, reliability, and recovery.
• Writing about AI product decisions, infrastructure tradeoffs, and the gap between compelling demos and dependable products.

• Skyflow Detect — AI privacy infrastructure for finding and protecting sensitive data across datasets, files, APIs, and agent workflows.
• JobAuto.ai — Founded and built browser agents that automated job applications across platforms including Greenhouse, Workday, and Ashby.
• PaperTalk — Built an AI product for working with research papers as part of my founder journey.
• Rakuten — Built and shipped automation and conversational-AI systems for global e-commerce before moving into AI product management.

• First author, An open-source data storage and visualization platform for collaborative qubit control, Scientific Reports (Nature Portfolio), 2024.
• Co-author, Computing systems for superconducting qubits: challenges and opportunities, ACM MobiSys, 2025.
• Former researcher at Lawrence Berkeley National Laboratory; M.S. in Computer Science from the University of Texas at Arlington.

• 0→1 AI Product Manager at Skyflow
• Former software engineer at Rakuten
• Berkeley Lab researcher and published first author
• Founder of JobAuto and PaperTalk

The best place to understand my work is devanshubrahmbhatt.com. You can also reach me through LinkedIn or at devanshu.vguj@gmail.com.