@Modular: We spent 90 minutes at Monday's community meeting watching developers demo the impressive projects they've built with M…

X AI KOLs Timeline Events
mojo community-meeting developer-demo reinforcement-learning async-runtime open-source

Summary

Modular held its Monday community meeting where developers demoed projects built with Mojo — an HTTP client, an RL/robotics system, a kernel packaging SDK, and an async runtime — including grant-supported work.

We spent 90 minutes at Monday's community meeting watching developers demo the impressive projects they've built with Mojo, including work supported by the Modular Community Grant Program: 1. Floki (Mikhail) is an HTTP client built on libcurl. Its API is close to Python's requests, so most existing code ports over with small edits. 2. Noira (Denis) covers reinforcement learning, a physics engine, and robot deployment. On an M1 Pro it solves LunarLander in 39 seconds. PyTorch + CleanRL takes 235. 3. Thyn (Angel) packages 6,000+ Mojo kernels as Python and TypeScript SDKs. 4. Warp (Alex) is an async runtime that works out when the CPU and GPU need to sync, which lets kernels run concurrently. Check out the recording:
Original Article
View Cached Full Text

Cached at: 10/01/26, 08:23 AM

We spent 90 minutes at Monday’s community meeting watching developers demo the impressive projects they’ve built with Mojo, including work supported by the Modular Community Grant Program:

  1. Floki (Mikhail) is an HTTP client built on libcurl. Its API is close to Python’s requests, so most existing code ports over with small edits.

  2. Noira (Denis) covers reinforcement learning, a physics engine, and robot deployment. On an M1 Pro it solves LunarLander in 39 seconds. PyTorch + CleanRL takes 235.

  3. Thyn (Angel) packages 6,000+ Mojo kernels as Python and TypeScript SDKs.

  4. Warp (Alex) is an async runtime that works out when the CPU and GPU need to sync, which lets kernels run concurrently.

Check out the recording:

Similar Articles