Mar 28
Loading video player...
Building a Chatbot Service with Rust, WGPU, and Tokio - Ada Hieta at EuroRust 2025 This talk is an exploration of what’s currently possible in the Rust ecosystem for running machine learning inference locally – without relying on CUDA or PyTorch. We’ll walk through building a chatbot service powered by Rust, using WGPU for GPU-accelerated compute via WGSL shaders, and Tokio for serving responses asynchronously over an API. A key focus will be on bridging the asynchronous world of WGPU’s GPU command submission with Rust’s async ecosystem, especially Tokio. We’ll examine how to coordinate GPU-bound compute tasks with an async server runtime, ensuring responsiveness without blocking critical paths. Rather than presenting a production-ready ML stack, this talk is a hands-on look at what developers can achieve today with Rust’s graphics and async tooling. We’ll explore what works well, what doesn’t (yet), and where the ecosystem is headed – aiming to inspire curiosity, discussion, and contributions. Speaker Ada Hieta I'm a Senior Applied AI Engineer at Softlandia, with focus on Rust and practical applications of AI technologies. I have been using Rust since 2017, and I have experience in web development, image analysis, metrology software and industry automation. She/Her & friend of the Blåhaj -------------------- EuroRust 2026 – Get your ticket for EuroRust 2026 now and join us in Barcelona! 👉 https://eurorust.eu EuroRust is organized by Mainmatter, experts in Rust and distributed systems 👉 https://mainmatter.com/rust/