Machine Learning Meets Rust: Hands-on for Research Engineers

In this hands-on workshop, we will explore Rust’s potential in machine learning workflows, with a focus on performance, memory safety, and practical integration into research pipelines. Participants will use the Rust library Kalosm to experiment with local text generation using pre-trained transformer models, as well as embedding-based document retrieval and simple vector search.

To conclude, the workshop will include a live demo using Crane, a Rust-based CLI for local language model inference, and Candle, a pure Rust tensor library developed by Hugging Face for CPU-optimized execution. The demo will show how to run a compact transformer model locally and generate responses conditioned on user queries by retrieving relevant context from a lightweight vector database.

Audience: Research engineers and technically skilled researchers (familiar with ML pipelines and Python)
Duration: 120 minutes (includes a 10-minute break)
Equipment: Laptop, IDE (VSCode or Cursor recommended), Rust language (v1.86.0 or greater) installed

Goals:

1. Highlight how Rust can support efficient, small-scale ML applications for tasks such as context-aware generation and document indexing across platforms.
2. Provide participants with a functional example of how Rust enables streamlined, low-latency ML pipelines without relying on external runtimes such as Python or C++.

Presenter Bio

Marianne Goldin is a backend and systems software engineer based in Seattle, Washington, specializing in memory-safe systems programming with Rust, applied cryptography, and distributed networks. She advocates for clean coding practices, open-source mentorship, and inclusive technological education communities. As part of the Rust Foundation Community team, she co-edits the widely read This Week in Rust newsletter, helping to inform and connect the global Rust community.

Driven by her personal commitment to broadening access to technology, Goldin draws on her own non-traditional path into tech to support others. She is committed to flattening the computer science learning curve and dismantling systemic barriers to industry entry. She is recognized as an impassioned and committed mentor with Mentors in Tech and CodeDay, guiding students through practical projects and in their career development, aiming to reduce systemic barriers in the computer science education pipeline.

Goldin is the founder of the arts nonprofit Src Material, supporting emerging new media art practices informed by technological innovation. Together with a team of mentors, their work bridges technical research and artistic communities, exploring intersections between digital culture, speculative tech, and conceptual art.