Super Data Science: ML & AI Podcast with Jon Krohn 967: AI for the Physical World, with Samsara's Praveen Murugesan
25 snips
Feb 17, 2026 Praveen Murugesan, VP of Engineering at Samsara who leads AI and edge-to-cloud work for transportation and logistics. He talks about processing massive sensor and video datasets, edge AI that works offline in long-haul vehicles, tiny models and Video LLMs for non-experts, routing and quantum possibilities, and what Samsara looks for when hiring.
AI Snips
Chapters
Books
Transcript
Episode notes
In‑Cab Cameras Run Edge AI
- Praveen described in-cab safety cameras that run vision models and temporal reasoning on edge hardware to detect drowsiness and phone use.
- He emphasized running inference on constrained devices to provide immediate driver feedback.
Cloud Training, Edge Inference, And Model Management
- Samsara trains models in the cloud but performs inference on edge devices and manages model versions for deployment.
- The team experiments with tiny LLMs on edge for tasks like voice translation alongside proprietary vision models.
Design For Long Offline Periods
- Customers often operate off-grid for days, so Samsara designs solutions to gracefully degrade and operate without connectivity.
- Edge-first safety features ensure critical alerts still function when networks are unavailable.
