Technical Product Manager, AI Inference & Software

About Positron AI

Positron AI is building next-generation AI inference accelerators designed from the ground up for low-latency, high-throughput large language model inference. Our first-generation ASIC, Asimov, is a cutting-edge accelerator targeting frontier AI workloads, with additional generations already underway.
Role Overview

Positron AI is looking for a Technical Product Manager to own AI inference and software technical product planning end to end. In this role, you will be the person who translates where models and inference systems are heading into concrete, well-scoped requirements for our inference software stack, spanning model coverage, numerics, inference-engine features and modes, serving-stack capabilities, and our managed service.

This is a deeply technical planning role that sits at the intersection of engineering, go-to-market, and the broader inference ecosystem. You will track the model frontier as a discipline, convert that movement into engineering requests before it becomes a customer escalation, and serve as the connective tissue between our engineering organization, our GTM teams, and our ecosystem partners. You will also be expected to use agentic AI daily as a core part of how the planning function operates.
Key Responsibilities

- Be the leader in Positron for all aspects of AI Inference & Software Technical Product Planning.

- Write requirements for our inference software stack spanning model coverage, numerics, inference-engine features/modes, serving-stack features/modes, and managed-service capabilities.

- Keep our planning ahead of where models and inference systems are going - track the model frontier as a discipline and convert movement into requirements before it becomes a customer escalation.

- Work with engineering and GTM on communicating a roadmap.

- Work closely with GTM teams to understand customer and market needs.

- Define and manage the software product lifecycle: versions and release trains, feature modes, model catalog and deprecation policy.

- Creative competitive briefings.

- Work with our key ecosystem partners - model labs, open-source runtimes, serving and orchestration partners - to understand their technology roadmaps.

- Create scope/feasibility frameworks to convert model and inference-system innovations into tangible engineering requests.

- Ensure our software products are well documented.

- Streamline & automate the product planning processes using agentic AI.

Required Qualifications

- 10+ years experience with a strong mix of these types of technical skills:

-
- ML systems, inference infrastructure, or serving-stack engineering with direct ownership of performance or architecture trade-offs - transformer internals at the operator level: attention variants, MoE routing, KV-cache mechanics, quantization formats and their hardware implications.

- Production inference serving at scale touching on multi-tenancy, latency SLAs (TTFT/TPOT), batching and scheduling, disaggregated serving, KV-cache management, and observability.

- The open-source inference ecosystem - runtimes (vLLM/SGLang-class), kernels, model ingestion, and how models are released, quantized, and adopted in practice.

- Performance analysis spanning models & systems: utilization reasoning, tokens per $ and per W arithmetic, benchmark design - able to build and defend the math personally.

- Competitive landscaping and analysis of inference providers and serving stacks, at a model lab, inference API provider, or AI hardware company.

- 10+ years with a strong mix of these type of personal & analytic skills:

-
- Quick learner with the ability to span model-architecture details up to fleet-scale serving systems. Stays current with the model and inference landscape.

- Excellent communication and people skills, comfortable navigating uncertainty, and driving a process of idea & decision socialization.

- Comfortable being the most technically grounded person in a GTM room and the mos