Isan Sahoo

I’m a Principal Member of Technical Staff at Oracle. I spent most of my career in Oracle Cloud Infrastructure building and operating control-plane systems for OCI’s global bare-metal and GPU fleet, and recently moved over to Oracle Health. The work I’m best known for sits on the operational side of cloud: fleet reliability, capacity recycling, and the unit economics that determine whether AI workloads on a hyperscaler are actually profitable.

One system I designed, the Smart Recycling Framework, is credited with $2.9M in annual savings in Oracle’s FY25 Cost of Cloud Report. Another, GPU Quick Recycle, shipped as an official Oracle product and has run 305,000+ operations so far at a 0.00% failure rate.

Outside Oracle, I’m a Fellow of the BCS and IETE, a CES 2026 Innovation Awards judge, an author or co-author of 20 IEEE-indexed conference papers, and a regular reviewer for IEEE and Elsevier journals, with 300+ verified reviews to date.

https://ieeexplore.ieee.org/search/searchresult.jsp?newsearch=true&queryText=isan%20sahoo

My most recent open-source contribution is a January 2026 LF AI & Data community blog article, co-authored with an engineer at Apple, on Budgets as Code for multi-tenant GPU clusters using Flyte, OpenLineage, and Marquez.