Karthi Sadasivan

If someone asked me to step into a car with no driver, do you know what my response would be? “ABSOLUTELY!”   This past week, my family and I had our first Waymo experience and it did not disappoint. Having personally worked on self-driving cars while I was at Stanford, I have to say that the technology has certainly advanced in a positive way. The user experience was mostly great (2 pieces of feedback below) and the safety aspects were outstanding. The Waymo that we were in was a Jaguar equipped with 2 screens: one for the front passenger, and one for the rear passengers so we could all see what the car was seeing. The Waymo was able to detect people walking, skateboarding, and scootering at a much farther range than a human could. It even could detect and highlight other Waymos! The car provided clear prompts on wearing seatbelts and how to unlock and open the door to get out. The Waymo obeyed all traffic lights and signs. It yielded to pedestrians and knew when it was safe to make a left while yielding to the other drivers that were going straight.   We only had 2 critiques. The first was that it was not very intuitive to find the pickup spot from the Waymo mobile app. We had to walk about 400 feet around a corner. Fortunately, we had about 5 minutes to find our car and it had our initials spinning on the top of the locked car. There is something satisfying about unlocking the doors with the app. The second critique was that as we were approaching our destination, we were prompted with a few additional nearby drop off location options, so we could change for a cost. It would have been better if the initial drop off location was as close to the desired destination as possible.   All in all, we really enjoyed the experience and used the service 2 more times on our trip! Excellent work Dragomir Anguelov, Eric Munoz, Srikanth Thirumalai, Mizuki McGrath, Satish Jeyachandran, Omer Baror, Rafael Aldana, Cedric Dupont, Vinutha Kallem, Nick Webb, and everyone at Waymo. Bring the technology to Houston. We’re ready!   Watch the short video to hear our raw and honest amazement from our first autonomous ride!

12

2 Comments

🎉 Major MapLibre success - Rivian integrates a commercial map data source (Google) with the Open Source map visualization stack (MapLibre), offering flexibility and styling consistency across multiple platforms, while avoiding vendor lock-in.

54

5 Comments

While the latest tools for GenAI are grabbing headlines, it's an open-source initiative that's making robust observability possible: OpenTelemetry. 🚀 In a world of increasing agency and non-determinism, we can't afford a black box. OpenTelemetry provides the standardized traces, metrics, and logs that give us crucial visibility into LLM-powered applications. This helps us to: Instrument every operation in a multi-step chain; Capture granular data like prompt/completion tokens, model parameters, and user IDs using standardized semantic conventions, enabling precise root-cause analysis; Monitor model-specific metrics like token usage and API call latency .. It’s why leading tools like MLflow, Arize, and Comet ML's Opik are all leveraging it in the backend. The power of OpenTelemetry is a testament to the open-source community's ability to solve the industry's most critical challenges. Check out the code and options to contribute here: https://lnkd.in/g6A_Acfy #GenAI #LLM #OpenTelemetry #Observability #MLOps #opensource #AI

15