ShipTalk - SRE, DevOps, Platform Engineering, Software Delivery · By Harness

From Windows Kernel to AI-Native Platforms: Scaling Engineering Teams and Systems with Puneet Saraswat

·41 min·2 clips

Puneet says "you can't do things without AI anymore" and ties it to agentic SDLC workflows.

1. ShipTalk - SRE, DevOps, Platform Engineering, Software Delivery turns into a career-and-systems conversation with Puneet Saraswat about Windows, Azure, Harness, and AI-native platforms. 2. Puneet Saraswat is the VP of Engineering at Harness, and Devan introduces him as the season three finale guest because he leads platform engineering and SRE work. 3. The episode asks why production systems fail at scale and how AI can help software delivery without weakening reliability or security. 4. Puneet says his interest in programming started in seventh or eighth grade when a school computer lab in UP, India, became a place to escape the summer heat. 5. He recalls building a BASIC report-card system with a friend as an eighth-grade summer project, which gave him fast feedback from code. 6. At IIT Kanpur, he studied computer science and worked on AI, machine learning, feature extraction, and computer vision more than 20 years ago. 7. He describes using MATLAB, C, OpenGL, and vision libraries for background segmentation and moving-body-part image segmentation. 8. Puneet says those AI and machine-learning topics were not yet popular in public culture, but they were present in academia. 9. Microsoft was his first job, and he joined a team building Windows 7 after the Windows Vista project ended and teams were reshuffled. 10. He spent his first three to four years on Windows kernel networking device drivers and learned production-grade systems from the Windows release cycle. 11. He says Windows had telemetry for bug checks and blue screens, which routed failures back to the right developer for root-cause analysis. 12. Puneet describes the lesson from that period as "there is no corner cases" when software runs at the scale of Windows. 13. When he moved to Azure in 2013 or 2014, he had to unlearn Windows-era assumptions and relearn distributed systems and continuous deployment. 14. He explains that Azure work pushed him to think about limiting blast radius by keeping traffic and data within a region. 15. He says synchronous APIs blocked threads, so the team introduced queues and workers to let capacity scale with outstanding requests. 16. Puneet says he joined Harness in January 2018 after Rishi, the CTO at the time, described a deployment problem that matched his Azure experience. 17. He says Harness had around 15 engineers and a few beta customers when he joined, and the company was building CD as its first product. 18. He now runs platform engineering and SRE at Harness, and much of his work is strategy, team alignment, and choosing projects that move the business forward. 19. He says he would hire generalists with strong analytical skills and a growth mindset, because Harness changes quickly and the business keeps evolving. 20. He would not skip the episode if he wants practical views on Windows, Azure, CI/CD, AI agents, or engineering leadership, and he might skip it if he wants a purely product-marketing discussion.

As heard by us

A plainspoken account of toil, scale, and the decision to build against it.

Ship Talk closes season three on a clear note, with Puneet Saraswat, Harness's VP of Engineering, reflecting on a path that runs from Azure's Application Gateway team to the operational friction that later made Harness feel necessary.

Read the full review in PlayNext →

Why you'd press play

You know the cadence is real when the next release is already on deck before the rollback settles.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from ShipTalk - SRE, DevOps, Platform Engineering, Software Delivery