๐ Kodaira, Japan ๐ฏ๐ต
As an AI Compiler Engineer on the Renesas HPC team, you will be responsible for driving compiler and code generation technologies that unlock the full compute potential of Renesas next-generation automotive System-on-Chip platforms, including advanced 3 nm silicon for software-defined vehicles (SDVs). Your work will directly impact how AI workloads ะ ัั from perception and sensor fusion to in-vehicle assistants and advanced driver assistance ะ ัั are translated into highly optimized, safe, and power-efficient execution on Renesas hardware.
This role bridges software compiler development, AI model lowering/optimization, and hardware-software co-design, enabling Renesas SoCs to deliver industry-competitive performance, efficiency, and functional safety required by multi-domain automotive applications..
ัััJob Descriptionััะ
ัะโLead AI compiler architecture across model ingestion, graph optimization, lowering, code generation, and runtime integration
ัะโDesign and implement graphะ ัะlevel optimizations (operator fusion, quantizationะ ัะaware rewrites, memoryะ ัะaware scheduling, partitioning)
ัะโDrive performance optimization for target NPUs, including tiling, tensor layout, and multiะ ัะcore execution strategies
ัะโPartner with SoC and AI accelerator architects to influence hardware features through compiler insights
ัะโOwn performance KPIs for real automotive AI workloads using simulators, profilers, and siliconะ ัะcorrelated models
ัะโEnsure compiler outputs meet automotive requirements (realะ ัะtime behavior, determinism, quality expectations)
ัะโMentor senior engineers and set technical direction without peopleะ ัะmanagement responsibilities
ัััImpact & Differentiatorsััะ
Renesasะ ัะ Gen5 R-Car automotive SoC lineup, including flagship 3 nm devices like the R-Car X5H, is among the first highly integrated multi-domain automotive SoCs built on advanced 3 nm process technology, designed to run ADAS, IVI, gateway, and next-gen SDV workloads on a centralized platform. These platforms deliver high AI performance (e.g., multi-hundreds of TOPS), scalable chiplet-based acceleration, and power efficiency tailored for electrified and autonomous vehicles while meeting stringent functional safety standards. An AI Compiler Engineer enables this hardware vision by ensuring that state-of-the-art AI models and computational kernels are efficiently mapped to the silicon fabric ะ ัั directly enhancing performance, reducing latency and energy, and accelerating software adoption in automotive ecosystems where compute efficiency and safety are paramount.
ัััMust-Haveััะ
ัะโMS/PhD (or equivalent experience) in Computer Science, EE, or related field
ัะโDeep experience building AI compilers, accelerator backends, or graph optimization frameworks
ัะโStrong expertise in graph optimization and performance optimization for NPUs or custom accelerators
ัะโExperience with MLIR, LLVM, TVMะ ัะlike systems, or proprietary compiler IRs
ัะโExcellent C/C++ and Python skills
ัะโSolid understanding of AI inference workloads (CNNs, transformers, perception or generative models)
ัะโStrong communication skills are required, e.g. agile development experience in Scrum team (Scrum Master or member)
ัััNice-to-Haveััะ
ัะโExperience with automotive or safetyะ ัะcritical systems
ัะโBackground in heterogeneous SoCs (CPU/GPU/DSP/NPU)
ัะโPerformance modeling or hardwareะ ัะsoftware coะ ัะdesign experience
ัะะคัะะััั
ััโฃัะยปััะััั
To Make Our Lives Easier
ััะโโัะกโโััะัะยซะขัยซััะ
ัะะัััะขะฆะนะ ัััะะััยปะ ัะฎัะะคัะะััะโโะ
ัะะตัะััะัPurposeัะยซะกะะััะัั
ััะโะฃะโัะโัะััโัะกะนะััะนัะััะะฆัะโััะธัะะัะโัััะขะัะกะะัะะัะะัะะัััะถะคััะะะฃโััะะตััะธััโฃัะััะะฐัะะณัะะัะะคัะัััะดัะะััะดััััะโัะะััะัะััะะชัั
ััะโะฃะโัะโัะััโัะกะนะัะยซัะััะโัะััะโัะะตัะะัะะดััะะฃะััะะะฃโัััะััะะขะฆะััะััััะโัะะัะะัะโIoTัััะถะัััะัะะัะะคััะัะะััััะะัะะััะัะโัะััะโััโฃััโัะโัะััะะฆัะโัะัััะัะโััโััะัั
ััะโะฃะโัะโัะะัะะััโัะะััะธัะโััโััะััะฑัะััะะััโโโัััโัะะััยปัะัััะัะะัะัััะััะัะะฎัะะัะะดัะะัะยปัะโัััััะคััะัะะชัโฃะัโะัะัะฃะะนัะะัะะฎัะโัะััะะััะัะัััััััะฃโะัะะตัะะัะะชััะััโฃััโัะโัะะัะััะะคัะะััะะขะะณััััะัััะนัะััะะฆัะโััะธัะะัะโัััะขะัะกะะัะะัะะดัะััะะัะะััั
ัะะคัะะััั
ััโฃัะยปััะ30ัะะัะะนะกโะฆะกะััะะ22,000 ะกโโัััะฃะฅะัะัััะัััะขะัะขัะัะััะั
ัััััะัะะะขะฆะัะะัะะตัะโัะะคััะะถะััะััะะคะขัะะขัะดัะะัะััะัััะ
ััะัะะััะััโัะะัััััโัะโััะธัะะัะโัััะถัััะััะะดัะะดัะโััะฅััะตััยปััโฃัะััะัััะตัะโััโฃััััะะัยซะชััััะะัะะััะะขัโัะะชัะัััััะััะะัะโััะธัะะัะโัะยซะขะัะกโะัััััะัััะถัะัะััะะดัะััะะัะะัะะััััะะฎัะะัะะดะกะััะัะกะะัะยซะกโโััะัััััโัะะชัะะฆัะะัะัััะัะยซะขััะขะฎะฆัะยซัะะชััะัะะคััะะขัะัะฅััะยปะฃะะนัะะััะััะตัะะััโฃะขัััะยซะถะคััะัััะนัะััะะฆัะโััะธัะะัะโัะยซะถัะัะโัะะคัะะตััะัะะัะัััััั
ััะโััะััั
To Make Our Lives Easier
ััะััััยซะชัะะัะะัะะัะะััั
ัะะคัะะััั
ััโฃัะะัยซะชัะะัะะัะะััะัะะัะะต
ะฃะัััััะยซััะัะะะขััััะชัััะขัโคัะโััะััะัะะัะัััะฑััััยปะ
ัะะะขโัััะัะยปัะะัะะัะะดัะััะะัะะัะะโโะช
ัะะคัะะััั
ััโฃัะะะกะััะธััะะค
ะขััะขะฎะฆััััะนะฑัะะฆัะะัะะัะะดัะััะะัะะัะะััะัะั
ััั
ัะนะััะัะะัะยปััะัะะััััะััะััะะัะะ
ัะััะะัะะฑัะะัะะคัััะขัะััะตัะะัะะดัะััััััะัะะะขะฆะัะะัะยปะถัโัะะค2ะขะะฆะถัะัะััะะฑัะโัะััะยปัะโััยปัััะฃะััะััะะัะะตัะััะะัะะัะะัะะััััััะขะััะะคััะะขยซะััััะยซะขะะฆัะยปัะะัะโัะะฐัะะตัะะัะะดััััะะััะััโฃัะะคะถะััะะัััััะัะััะะััััโะธััััะะัะะดัะััะะัะะััััะโััะะขัะัยซัะขะะฆัะยปัะะคะขะัะขะะฆัะะััะ
ะขัะตะขะัะขะะฆัะะััะััััะััะะัะโััะธัะะัะโััะััโัะะัะััะะณัะโััะธัะะัะโััะัะะฎัะะัะะดัะฅะัะฅัััััะััะะดัโััะะคัะัััััั
ััััะขะะฆัะะตัะะัะะดัะััะะัะะััั
Free forever • No spam • Leave anytime