Super Data Science: ML & AI Podcast with Jon Krohn · Jon Krohn

967: AI for the Physical World, with Samsara's Praveen Murugesan

·55 min·3 clips

Commercial navigation changes routes for a truck with a 20-plus-foot trailer and can enforce company-specific turn policies.

The episode begins with a hard operational problem. Jon Krohn frames it through a truck driver asleep at the wheel and the reality of miles without cell signal, which quickly shows why AI at the edge matters for physical systems. Praveen Murugesan joins from Amsterdam. As VP of Engineering at Samsara, he describes a company that supports construction, transportation, manufacturing, retail, logistics, and the public sector while processing more than 20 trillion data points. That scale shapes the whole conversation. The discussion turns to how Samsara makes AI useful across physical applications, which means treating deployment and enterprise readiness as part of the product rather than an afterthought. A gateway sits at the center. Praveen explains that Samsara built a layer of intelligence around model access so teams can use OpenAI, Anthropic, and Gemini while keeping cost and internal rigor around security and compliance in check. Engineers do not have to carry all of that themselves. Instead, the platform lets data scientists and engineers hand off those concerns to another team and get the capability out of the box. Feedback loops matter too. Praveen says the setup is meant to work like a training ground before anything goes broadly into production, giving teams fast signals about cost profiles, quality, and promotion paths. The show then lands on a concrete customer problem. A product engineer inside the organization built a solution for a common request: identifying when drivers are misusing the vehicle assigned to them, which historically meant having to sift through all the content. That example keeps the episode grounded. It shows how a platform team can turn an abstract AI capability into something tied to policy, behavior, and day-to-day operations. The questions move from broad systems thinking into the mechanics of model choice, feedback loops, and enterprise guardrails, and Praveen keeps the answers close to what the platform actually does. The episode stays practical throughout. Listeners who want a clearer picture of how AI gets applied in fleets and other physical environments get a steady look at the tradeoffs, the tooling, and the operational consequences.

As heard by us

Edge AI is treated here as a practical tool for physical operations, not just a model decision.

Praveen Murugesan and Jon Krohn look at where AI actually has to hold up under pressure: places with shaky connectivity, messy conditions, and real work that cannot wait for a clean demo.

Read the full review in PlayNext →

Why you'd press play

You want a clear view of how edge AI turns physical-ops data into action.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Super Data Science: ML & AI Podcast with Jon Krohn