Principal Product Manager

SambaNova is a leader in next-generation AI infrastructure, delivering a full-stack inference platform for customers worldwide. At the core of SambaNova’s technology is the RDU (Reconfigurable Dataflow Unit) — a chip built on a dataflow architecture rather than the traditional GPU model. Its decode performance is especially strong for agentic workloads like multi-turn agents, code generation, and long-running applications. RDUs are packaged into SambaRack, rack-scale hardware that lets customers deploy state-of-the-art models with better performance, greater energy efficiency, and faster time to value.

We are looking for a technical product manager to own SambaNova’s model and capability roadmap end to end. You will be the product owner and the subject-matter expert on models: you understand why a model matters (architecture, benchmarks, adoption, customer pull) and what it takes to ship it well (function calling, structured output, reasoning, long context, caching, speculative decoding and more). Most importantly, you turn that understanding into decisions, commitments and launches. This is a product role for someone deeply technical, not an engineering role with product tasks attached, but you will stay hands-on: testing new models yourself, finding where they shine and where they break, and building quick mock-ups that show the competitive edge fast inference unlocks.

Responsibilities

Own and drive the roadmap

• Model Stack Rank: Own the unified, weighted stack rank of candidate models across customer demand, capability, market adoption, recency, strategic value and engineering effort. Tune the rubric, make and document overrides, and run the regular prioritization review
• Per-Chip Roadmap: Own the plan of models × hardware generation × configuration × performance targets, and keep it grounded in measured platform data rather than estimates
• Commitments and Launches: Turn priorities into dated commitments with engineering, track them to delivery and run model launches with Marketing, Docs and GTM
• Roadmap Communication: Keep the model roadmap and dashboard current as the source of truth for execs, account teams and customers, and present it with confidence both internally and externally

Define the capability bar

• Feature-Completeness Framework: Own the definition of a “fully featured” model across speed, efficiency, model infrastructure, cloud and API features, for example tool calling, JSON schema and constrained decoding, reasoning controls, full context length, prompt caching, speculative decoding, streaming, logprobs and API compatibility (OpenAI Chat, Responses, Anthropic)
• Gap Management: Find capability and conformance gaps from testing and telemetry, set their priority and drive them to closure with engineering
• Quality Bar: Work with QA and Solutions to set acceptance criteria and the evals a model must pass before it launches

Get hands-on with models

• Model Testing and Evaluation: Get your hands on new models as soon as they land: run them through real prompts, agentic tasks and our feature checks to learn what works well, what does not, and where the gaps are
• Finding the Edge: Spot the models, capabilities and use cases where SambaNova’s speed and efficiency create a real competitive advantage, and back that view with evidence
• Prototypes and Mock-ups: Build quick demos, prototypes and product mock-ups (e.g. a small app, notebook or UI) that make a model’s value concrete for leadership, GTM and customers, and hand the strongest ideas to Solutions Engineering to productionize
• Hands-on Findings to Roadmap: Feed what you learn into the stack rank, the feature bar and launch messaging, so roadmap decisions draw on first-hand experience as well as the data

Be the model SME

• Market and Model Intelligence: Track the open-weight model landscape (new releases, architectures such as MoE, MLA and GQA, benchmark and usage signals from sources like Artificial Analysis, OpenRouter and Hugging Face) and form a clear view on what matters and why
• Customer Voice: Take in requests from customers, the field and the Model Feedback process, triage them, close the loop with requesters, and make sure demand is represented accurately in the rank
• Competitive Positioning: Understand how our models, speed and features compare with GPU clouds and other inference providers, and turn that into roadmap and messa…