Positron
Modeling & Simulation Engineer
$200K–$300K
Remote (United States) · Remote · Full-time · Posted Aug 3, 2026
- Education
- BS, MS, or PHD in CS, EE (or related field)
- Visa sponsorship
- Candidates must be currently authorized to work in the U.S.; no new visa sponsorship is available, but H-1B visa transfers may be facilitated for eligible candidates.
About the job
Build and validate simulation infrastructure and performance models for AI inference accelerators, enabling software development and architecture decisions before silicon is available.
What you'll do
- Develop analytical, functional, and timing models of inference workloads and accelerator components.
- Integrate simulators with virtual platforms and production-aligned firmware, runtime, and host software; improve simulation speed, usability, and test coverage.
- Run reproducible architecture experiments, validate model behavior, and communicate assumptions, limitations, and tradeoffs.
What they ask for
- Strong systems-language software engineering skills, such as C++ or Rust, and experience building simulators, performance models, emulators, or comparable tools.
- Deep understanding of processors, memory hierarchies, interconnects, concurrency, and performance effects of data movement and contention.
- Experience designing and validating models for fidelity, determinism, and speed; strong applied mathematics and quantitative reasoning.
- Ability to integrate and debug complex software systems, build maintainable tools, and communicate technical assumptions and tradeoffs clearly.
Nice to have
- Experience with SystemC, Arm Fast Models (LISA), QEMU, gem5, or custom simulation frameworks.
- Experience modeling AI inference workloads, or familiarity with embedded Linux, firmware/runtime integration, device models, PCIe, or direct memory access.
Pay
$200,000 – $300,000 USD base salary
Tools
C++, Rust, SystemC, Arm Fast Models (LISA), QEMU, gem5, Haskell, Lean
AI focus
Model AI inference workloads and accelerator performance, including latency, throughput, utilization, tensor operations, quantization, memory traffic, and token-generation performance.
Model-written summary · the original posting is authoritative