Baseten is scaling fast
Baseten is visibly ramping growth — an SF billboard campaign plus a new Post‑Training Research Engineer role point to a push into post‑training/model‑ops and demand generation. This positions them squarely in the emerging ‘agentic AI’ application layer that NVIDIA highlighted at GTC, meaning they’ll likely prioritize low‑latency, high‑throughput infra as they scale. (techspot.com) (hyperight.com) (linkedin.com)
Baseten’s new Post‑Training Research Scientist listing explicitly tasks candidates with designing and running large‑scale experiments on multi‑node, 1T+‑parameter models and publishing results at top venues like NeurIPS/ICLR. (jobgether.com)) The company’s website promotes an “Inference Stack” and pre‑optimized Model APIs that include entries for GPU‑oriented runtimes such as NVIDIA Nemotron 3 Super, signaling an emphasis on GPU‑optimized production paths. (baseten.co)) Baseten’s careers and jobs pages name customers including Cursor, Notion, Gamma and Writer and quote that its Forward Deployed Engineers helped scale a partner to “over 70 million users and billions of requests.” (jobs.ashbyhq.com)) The firm is hiring across model performance and infrastructure teams — open roles explicitly list GPU kernels, GPU networking & distributed systems, and model performance engineering positions. (jobs.ashbyhq.com)) Baseten has recently pushed beyond inference: VentureBeat reported a general availability launch of “Baseten Training” as the company moves into training/post‑training offerings, and BusinessWire recorded a $75M round in February 2025 to fund product and geographic expansion. (develop.venturebeat.com)) San Francisco billboard data show outdoor ad rental growth of roughly 30% (2023–2025), and Baseten’s head of marketing Mike Bilodeau said the cryptic AI ads are “quite clearly not advertising to the average consumer,” underscoring a demand‑generation play targeted at builders. (techspot.com))
