Developer Relations Engineer

This is our first DevRel hire, and it is a content and developer-experience role. You’ll own the material that teaches researchers and training engineers how to get real work done with Spiral: cookbooks, worked examples, benchmarks, and reference pipelines grounded in datasets people actually train on. You might also conduct research on new applications of the core Spiral product.

You’ll collaborate with the rest of the team to create the tightest possible feedback loop between developers/researchers and our product and docs. The job is to earn credibility with a technical audience by shipping things that are genuinely good.

What you’ll do

• Write technical cookbooks and end-to-end examples - video ingestion, GPU data loading for training, multimodal feature engineering - that a researcher can run and immediately understand.
• Build and maintain reference pipelines against real datasets (such as those on Hugging Face), and keep them working as the product moves.
• Produce credible benchmarks and the honest writeups that go with them.
• Be the developer’s advocate internally: turn friction you and users hit into concrete product, client SDKs (e.g. pyspiral), and docs improvements.
• Answer real questions in the places our users already are (GitHub, community channels, conferences), and turn recurring ones into permanent docs.
• Represent Spiral at conferences.
• Give the occasional talk or workshop.

• You’ve done ML or training-infrastructure work yourself, and you’ve personally felt the data-loading / video-decode / GPU-utilization pain we remove.
• You write well about technical things. You can point us to things you’ve written.
• You’re fluent in Python and comfortable in the PyTorch / Hugging Face / data-pipeline ecosystem.
• You have a working mental model of GPUs, training loops, and where the bottlenecks actually are.
• You are well connected in developer and/or AI communities.

Nice to have

• Open-source contributions in relevant territory (PyTorch data / DataLoader, HF datasets, Ray Data, video/decode tooling, or similar).
• Experience with columnar / analytical data formats.
• An existing audience among ML practitioners - welcome, but genuinely secondary to the credibility above.
• Prior DevRel, AI research, or research-engineering experience.