F
Fractile Tech Stack
AI inference accelerator chips and system software for large language models
Fractile designs custom silicon and systems software purpose-built for AI inference at scale. The stack reflects a full-stack hardware play: SystemVerilog, MLIR, PyTorch, vLLM, plus low-level tools (Cadence Innovus, Synopsys ICC2, Calibre) for chip design and verification. The company is actively adopting Rust and SystemC while hiring heavily in engineering (senior-weighted, UK and Taiwan), and their project list centers on pre-silicon modeling, RTL specification, Linux kernel drivers, and data-center deployment—indicating they are moving from design into manufacturing and production scaling.
Tech Stack 50 technologies
Core StackPython Rust Docker Kubernetes PyTorch PCIe Gen 6 PCIe 5.0 JTAG SystemVerilog Cocotb Verilator Bazel UVM SystemC C++23 ARM x86-64 RISC-V MLIR C/C++ vLLM SGLang QEMU Linux Kernel Apple Silicon Cadence Innovus Synopsys ICC2 Calibre TCL Perl+20 more
What Fractile Is Building
◆Challenges
- High-volume manufacturing
- Deploying ai accelerators into data centres
- Delivering product at scale
- Meeting demand for high-performance ai silicon
- Accelerating large language models
- Scaling production with manufacturing partners
- Yield ramp acceleration
- Meeting demand and scaling product delivery
- Improving tape-out quality
- Bringing software concepts to hardware workflows
▲Active Projects
- Develop systems software for ai accelerator hardware
- Develop and maintain rtl specifications for high-performance systems
- Launch ai accelerators into data centre deployment
- Pre-silicon modeling of ai acceleration hardware
- Early software development using models of products in development
- Runtime stack for ai accelerators
- Ai accelerator deployment
- Linux kernel driver in rust
- Ai accelerator chip test solutions
- Dft architecture for large multi-core ai accelerator soc
Hiring Activity
50 roles · 25 in 30d
Department
Seniority
Company intelligence
