Senior GPU Software Architect

Senior GPU Software ArchitectBangalore / HyderabadFounded by highly respected Silicon Valley veterans - with its design centers established in Santa Clara, California. / Hyderabad/ BangaloreOur pay comprehensively beats "ALL" Semiconductor product players in the Indian market.Senior GPU Software ArchitectLocation: Hyderabad, or Bengaluru - IndiaDepartment: GPU Architecture & Software SystemsExperience Level: 15+ years in GPU, AI/ML, and Compiler ArchitectureJob SummaryAs a Senior GPU Software Architect, you will lead the design and development of the complete software ecosystem for our novel GPU backend. This includes compilers, drivers, runtimes, cloud integration, and AI frameworks designed to support next-generation AI workloads and graphics. You will work closely with hardware teams to co-create novel solutions while driving innovation that results in foundational IP and patents.---Key Responsibilities· Lead the end-to-end architectural design and implementation of the GPU software stack, from compilers to drivers to runtimes, for AI training/inference and advanced graphics.· Collaborate intensively with hardware architects and engineers to execute hardware-software co-design strategies that maximize performance and efficiency.· Create scalable and flexible software frameworks for emerging AI workloads, including LLMs, foundation models, and multi-GPU training clusters.· Architect GPU drivers supporting modern graphics APIs, ray tracing, and ML acceleration with a startup’s agility and focus on quality and innovation.· Pioneer customized compiler and runtime solutions leveraging LLVM/MLIR, Triton, TVM, Pytorch or other open-source frameworks optimized for our GPU architecture.· Design multi-GPU and cloud-native frameworks optimized for containerized AI training and inference workloads.· Drive innovative patent creation and develop novel software/hardware interaction techniques that differentiate our product in a competitive market.· Mentor a passionate, early-stage engineering team establishing a culture of technical excellence and ownership.---Required Qualifications· Solid 15+ years in GPU software architecture or AI accelerator software development, ideally with startup or product innovation experience.· Expert-level understanding of graphics pipelines, ray tracing, and AI workloads including LLMs, diffusion models, and tensor computations.· Hands-on experience with compiler infrastructure (LLVM, MLIR, Triton, TVM, PyTorch) and GPU code generation.· Practical experience in GPU driver development covering user-space and kernel-space components.· Strong hardware knowledge of GPU compute units, memory hierarchy, interconnects (NVLink, PCIe), and multi-GPU scaling.· Experience with GPU programming models (CUDA, HIP, SYCL) and AI frameworks (PyTorch, TensorFlow).· Knowledge of cloud-native architecture for GPU workloads, including container orchestration, virtualization, and multi-tenant GPU sharing.· Demonstrated innovation through patents or research publications in GPU/AI software or hardware/software co-design.· Proficiency in C++, Python, and system-level programming.---Preferred Qualifications· Experience building GPU software stacks in small/agile teams or startups.· Familiarity with low-precision numeric formats (MXFP4, FP8, BF16) and their use in ML workloads.· Background in heterogeneous compute environments combining CPUs, GPUs, and custom accelerators.· Prior involvement with GPU bring-up, simulator development, or performance profiling.· Advanced degrees (Master's/Ph.D.) in Computer Engineering, Computer Science, or Electrical Engineering.---What You’ll Gain· The opportunity to architect the foundational software for a game-changing GPU product aimed at disrupting the industry leaders.· Hands-on collaboration with visionary hardware and software engineers to build a world-leading AI GPU platform.· Ownership of key innovations, patents, and technical roadmaps shaping the future of GPU computing.· A pioneering startup environment that empowers bold decision-making, creativity, and rapid impact.Contact: Uday BhaskarMulya Technologies"Mining the Knowledge Community"Email id : muday_bhaskar@yahoo.com