SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Efficient Computer is building next-generation energy-efficient, general-purpose processors based on the Fabric architecture, a novel spatial dataflow design. The company's Electron E1 processor is shipping in volume, and they recently announced $97 million in Series B financing.
You will own the user-mode runtime components that interface with the Fabric architecture and support the effcc compiler. This is a critical role on the System Software team, requiring you to design, implement, and maintain the user-mode runtime driver for the Fabric accelerator—a shared library sitting between the compiler and the Fabric ABI. Your responsibilities include context management, program loading, resource management, command queue construction and submission, descriptor allocation, streaming, completion queue/event synchronization, error handling, logging, memory allocation, and general library support.
Key responsibilities:
- Design and implement all user-mode interfaces between the compiler, Fabric ABI, and kernel-mode driver
- Define in-memory representations generated by the compiler and consumed by the Fabric
- Implement runtime library support and interfaces for 3rd-party ML framework backends (PyTorch, ONNX, TensorFlow)
- Lead user-mode efforts for first-silicon bring-up, including work in functional simulation, emulation, and FPGA environments
- Establish comprehensive runtime test coverage: API unit tests, conformance tests, emulation-based regression suites, performance benchmarks, and hardware-in-the-loop validation
You will work closely with compiler, kernel driver, architecture, micro-architecture, and verification teams throughout all development phases. The role involves pre-silicon environments including QEMU, functional simulation, emulation, and FPGAs to ensure applications run on day one.
**Requirements:**
- B.S. in Electrical/Computer Engineering, Computer Science, or equivalent (M.S. a plus)
- 7+ years of systems software experience, including significant hands-on Linux user-mode driver and language runtime library development for accelerators (GPUs, NPUs, DSPs, FPGAs, or similar)
- Experience building or contributing to a production accelerator runtime software stack, from driver API through user-facing runtime library
- Solid understanding of accelerator memory models, particularly unified memory systems, allocators, and zero-copy data paths
- Experience designing public APIs for runtime libraries: ABI stability, versioning, error models, thread safety, and documentation
- Experience working with compiler teams on binary formats, loaders, launch ABIs, and kernel metadata for ahead-of-time or just-in-time compilation
- Experience integrating runtimes with language runtimes or ML frameworks: Python bindings, framework backends, execution providers, and C/C++ application interfaces
- Hands-on experience with QEMU, emulation, or simulation environments during pre-silicon development
- Experience designing kernel/user-space interfaces: ABI and HAL design, synchronization, memory sharing, and error handling
- Strong user-mode debugging skills: crash analysis, sanitizers, tracing, profiling, race and synchronization diagnosis
- Working knowledge of ARM/AArch64 Linux systems and the Linux driver model
- Expert C and C++; strong Python skills; comfortable with AI-assisted development tools
**Desired Qualifications:**
- Experience with CUDA Runtime (cudart), ROCm (HIP, ROCr, HSA), Tenstorrent runtime (TT-Metalium), oneAPI Level Zero, OpenCL, or similar accelerator runtime stacks
- Familiarity with LLVM or MLIR and how accelerator compilers hand off to runtimes
- Knowledge of dataflow, spatial, or other novel compute architectures and how compilers/runtimes target them
- Experience on first-generation silicon products where software and hardware matured together
- Experience with FPGA prototyping and hardware emulation platforms (Palladium, Zebu, Veloce, or similar)
