EP 76 — OpsLab's Arun Nair on Teaching AI the Art of Scheduling

OpsLab's scheduling platform saves Air Force squadrons 200 hours per month — time previously spent by pilots manually fixing flight schedules until late at night. Arun Nair, Founder & CEO, showcases how OpsLab's three-layer intelligence system went from 30% scheduler acceptance to operational deployment across 75+ squadrons by combining operations research algorithms, reinforcement learning from daily user feedback, and LLM interfaces that let commanders simulate scenarios like losing 20% of their fleet while managing ATOs and continuation training simultaneously.

Arun warns how Phase III SBIR contracts aren't the finish line everyone thinks they are — the real Holy Grail is program-of-record status. His team spent four years collecting feedback from schedulers to calibrate their algorithms, hired former Air Force schedulers to understand the dysfunction, and built a middleware layer borrowed from his Wall Street high-frequency trading experience that can combine air power from different branches onto the same platform. 

Topics discussed:

• Building AI scheduling architecture combining operations research, reinforcement learning from human feedback, and LLM interfaces.
• Achieving 200 hours monthly time savings per squadron by eliminating manual scheduling work that kept pilots working late nights.
• Scaling across 75+ squadrons and multiple aircraft types from single-seat fighters to B-52 bombers with eight crew positions.
• Understanding Phase III SBIR contracts are still R&D phase, not a sustainable business model, with program of record as the true goal.
• Hiring former Air Force schedulers to decode organizational dysfunction and collecting daily feedback to calibrate algorithms.
• Targeting 60 VCs for Series A fundraising, narrowing to 15 serious prospects, and using venture debt to extend runway.
• Applying Wall Street high-frequency trading middleware concepts to create a pub-sub layer combining air power from multiple military branches.
• Deploying LLM-powered scenario simulation enabling commanders to model fleet losses, ATOs, and continuation training constraints simultaneously.  

Listen to more episodes: 

Apple 

Spotify 

YouTube

Website