Back to search
About Vast.ai
Vast.ai runs one of the world's largest GPU marketplaces: 20,000+ GPUs, from RTX 4090s to B300s, serving 25,000+ monthly customers who train, fine-tune, and serve AI models. We're profitable, flat, and we ship fast. Applications go straight to the hiring team.
The role
This is an engineering role that happens in public — not a content-calendar job. You'll be Vast's resident power user: most of your week is spent renting GPUs on the platform and making them do impressive things — deploying and serving open-source models with vLLM, SGLang, PyTorch, and ComfyUI, keeping live endpoints running (including token endpoints on markets like OpenRouter), and building the example repos, templates, and benchmarks that show developers exactly how to do the same.
Then you teach it: short videos, technical guides, Discord and Reddit answers, hackathons. You won't be building the product — you'll be using it harder than any customer, in public, and feeding what breaks straight back to engineering.
What you'll do
Run real workloads on Vast weekly: deploy, fine-tune, and serve open-source models; keep live inference endpoints up
Build public example repos, Docker templates, and benchmarks; contribute to our open-source CLI/SDK and serverless tooling
Turn what you build into teaching: guides, short videos, talks, livestreams
Be the most technically credible voice in our Discord, Reddit, and X communities, and represent Vast at hackathons and conferences
Close the loop: report the friction, breakage, and missing docs you hit directly to engineering and product
What we need to see
You can take an open-source model from Hugging Face to a running endpoint on rented GPUs without hand-holding: Linux, Docker, Python, SSH
You can debug the GPU stack: driver/CUDA mismatches, OOM errors, and multi-GPU config (NCCL, topology) don't scare you
Public proof of shipped work: GitHub, technical writing, or live projects — we'll ask you to walk us through code you wrote
Self-directed: no one here will hand you a roadmap, and you prefer it that way
You can explain hard things clearly, in writing and in person
Nice to have
Prior DevRel, community, or client-facing experience (a bonus here, not the job)
vLLM or SGLang internals, distributed training (FSDP, DeepSpeed), CUDA
Time at a cloud, GPU, AI-infrastructure, or developer-tools company
Content or projects that found a real audience
First 90 days
Ship several public example projects on Vast, including at least one live serving endpoint
Publish benchmarks or guides developers actually use and share
Become the recognizable technical voice in our community channels
Represent Vast at a sponsored hackathon
Compensation & benefits
$160K-$200K base + equity + bonus. Comprehensive health, dental, vision, and life insurance; 401(k) with match; onsite meals; and a travel/conference budget. On-site in San Francisco or Los Angeles.
Learn more about this Employer on their Career Site
