Kai Wang on Inside Uber’s AI Revolution – Everything about how they use AI/ML. MLOps Podcast #329. 🎧 #MLInfrastructure #AIAtScale #MLStrategy #AIInProduction #UberTech #MLOps.

Kai walks us through how Uber built internal systems to tier ML projects, helping teams align on what’s critical and what’s experimental. It’s a story about scale, maturity, and knowing when to pivot your strategy.

By 2019, ML wasn’t the blocker anymore—performance was. That’s when the focus shifted: not just getting models into production, but making sure they were driving real value.

In this episode of the MLOps Podcast, Kai Wang shares the evolution of Uber’s ML journey—from enabling teams to adopt machine learning, to strategically tiering projects by business impact.

From Kickstarting ML at @Uber to Prioritizing What Matters Most. “When we started, there were just three machine learning use cases on our platform. Today, there are thousands.”

Shout-out to Philip Cannon and Nick Burka for helping make this stuff real. The work they’re doing is changing how advocacy happens—quietly and powerfully. 🎧 Catch the full episode here:

3️⃣ Keeping up with the rules:. Regulations are always shifting. AI helps make sure nothing slips through the cracks.

2️⃣ More time for people, not paperwork:.With the grunt work handled, attorneys can actually focus on their clients—not just the admin.

A few highlights from the chat:. 1️⃣ Sorting the chaos:.Social Security disability claims come with mountains of medical records. AI helps cut through the noise and surface what really matters.

Ever wonder what happens when AI meets legal red tape?. Nick Coleman joined Demetrios on the MLOps Community podcast to talk about how his team is using large language models to help attorneys actually spend more time advocating—and less time drowning in paperwork.

RT @MLflow: 📣 Happening This Week: Agents in Production 2025!. Join us virtually this Thursday, July 17, for "Driving Evaluation-Driven Dev….

Hit play here → and hear how it’s all unfolding.

3️⃣ Room to rise.As things localize, a bunch of new players are stepping into the spotlight. It’s a pretty exciting moment for anyone looking to build something new outside the usual hotspots.

2️⃣ Fast moves, smart builds.Who says data centers take forever? Some countries are flipping the script—think existing buildings + clever partnerships = AI-ready in months, not years.

Here’s a quick taste of what came up:. 1️⃣ Local is the new global.Frank broke down how more countries are keeping AI operations on home turf—data, models, all of it. Turns out, this shift is already shaking up the landscape far beyond the US.

Just dropped a new MLOps Community SF Live podcast episode with Frank Meehan—and wow, it’s a good one!. We got into how the global AI race is heating up, especially as countries start pulling their AI efforts in-house.

The Missing Data Stack for Physical AI // MLOps Podcast #328. 🎧 #PhysicalAI #AIInTheWild #RobotUX #AIInfrastructure #MLOps #RealTimeAI.

Speed isn’t just a UX preference—it’s a constraint of reality. Nikolaus West on time, responsiveness, and why planning in isolation isn’t enough in the physical world.

It’s not just about how long a robot takes—it’s about whether the action is still relevant once it finally moves. The world is dynamic. If the object’s gone, if the moment passed, it doesn’t matter how smart your agent is.

When Robots Miss the Moment. “Nobody wants a robot that comes back 20 minutes later and says, ‘Actually, I can’t do that.’”. In this episode of the MLOps Podcast, Nikolaus West gets to the core of what makes physical AI hard: the world doesn’t wait.