Senior Product Manager, Inference Platform

Job summary

Responsibilities

We're seeking a senior product leader to own Cohere's Inference as a Service business end to end. The product is sovereign, self-serve inference: a mid-market or enterprise buyer picks a model, a region and a GPU size, and is running the same afternoon with no procurement cycle and no…

• Own the P&L for Inference as a Service: carry the revenue number, set the rate card, and keep blended margin above the floor as the catalog and the fleet grow.
• Own the inference platform itself: the console and dashboards where customers see usage, spend, API keys, access control and attestation, and the value-add tooling that makes a…
• Own the model inventory across first-party and third-party models: what enters the catalog, how it is grouped by provenance, what it is priced at, and when something is…
• Set the roadmap and the investment case across serving efficiency, the self-serve surface and regional capacity, and make the call when those three compete for the same budget.
• Build the commercial motion with GTM from the ground up: self-serve acquisition for the mid-market, a sales-led motion for enterprise, and the partnerships that monetise capacity…

What they're looking for

• 10+ years of product management experience, including meaningful time owning a B2B infrastructure or platform product at the scope of a business rather than a feature area.
• Experience operating in a GM-like capacity: you have carried a revenue line or a P&L, set pricing, and been held to margin as well as to shipping.
• Background at a neocloud such as Together AI, Baseten or Fireworks, or at a hyperscaler such as AWS, Azure, OCI or Google Cloud, with a working view of how inference capacity is…
• A track record of thinking about infrastructure as a product, with the ability to translate system-level capabilities into clearly differentiated, customer-facing features.
• Entrepreneurial, with experience taking a new product to market and building the commercial motion around it rather than inheriting one.

Tech stack

• AWS
• Azure
• GCP
• LangChain
• LlamaIndex
• vLLM