TixelJobs
P
Positron AIvia muse

Performance Modeling Engineer, AI Inference Deployments (Senior/Principal)

REMOTE$220K - $270K/yrPosted 2d ago
ML EngineerStaff+Full-time#remote

About the Role

**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. Chips built for the training era keep running into two walls at inference time: memory bandwidth and chip\-to\-chip networking. We designed our architecture to break through both. Our first\-generation Atlas systems are already in production on Oracle Cloud Infrastructure. Asimov, our first custom ASIC, takes that architecture to silicon. In simulations, Asimov delivers several times more tokens per dollar and tokens per watt than leading GPUs, at a fraction of the power. Paired in our Titan system, Asimov is built to serve models beyond 16 trillion parameters. In September 2026, we raised $875 million at a $5 billion valuation. NEA led the round, joined by Atreides Management, Valor Equity Partners, and others, to bring Asimov and Titan to market. Our team includes engineers from Groq, Lambda, NVIDIA, Meta, F5, and many other industry leaders.

**Role Overview**
-----------------


As a Senior Performance Modeling Engineer, you will own the tools that tell customers how a model will run on Positron hardware at scale: how many chips it needs, what throughput and latency it will deliver, and how that compares with GPUs and other accelerators. You will work directly with customers, product, and our chip modeling team to turn deployment questions into quantified answers they can plan against.
The work spans analytical modeling, simulation, and real hardware measurements. You will build deployment\-level models on top of our chip performance simulators, validate them against hardware, and package the results so customers can run them on their own. We are hiring two engineers for this work: one will take primary responsibility for the modeling architecture, and the other for per\-model deployment studies and the customer\-facing tool.

**Key Responsibilities**
------------------------

**Customer and Product Partnership*** Partner with customers and product to turn deployment questions into focused studies with clear assumptions, options, metrics, and recommendations.
* Run apples\-to\-apples comparisons of the same deployment on competitor hardware and explain what drives the difference.
* Design and deliver a standalone performance\-estimation tool that customers can run on their own workloads.

**Deployment and System Modeling*** Model deployment topologies for large\-scale inference, including tensor, pipeline, expert, and data parallelism, batching, context length, and mixed Positron\-plus\-GPU configurations.
* Model the network and system layer, including interconnect latency and bandwidth, redundancy, and how these bound throughput as chip count grows.
* Develop and extend analytical and simulation\-based deployment models that consume our chip performance simulator, so every estimate traces back to one shared hardware model.

**Validation and Calibration*** Validate projections against measured hardware runs and keep models calibrated as chips, software, and workloads evolve.

**Required Qualifications**
---------------------------

* 5\+ years of experience in performance modeling, capacity planning, or systems architecture for large\-scale AI inference, HPC, or data\-center systems.
* Demonstrated experience building analytical or queueing performance models and calibrating them against measured systems.
* Deep understanding of LLM inference at the systems level, including prefill versus decode, KV cache, parallelism strategies, mixture\-of\-experts routing, batching, time\-to\-first\-token, and tokens per second.
* A solid working knowledge of data\-center networking (latency, bandwidth, and topology) and GPU performance characteristics.
* Comfort working from incomplete information, stating assumptions explicitly, and delivering an answer with error bars rather than waiting for certainty.
* Strong Rust skills, or expert\-level C/C\
Share