Pratikkumar Patel

Drive vision and strategies for Nvidia GPU and SoC performance, power, Thermal software roadmaps working with cross-functional HW, SW, Architecture, Modeling and QA teams. Mentor and guide organization of technical managers with 30+systems software engineers.

Collaborated with cross-functional teams to gather insights and conduct gap analysis, driving innovation in Nvidia GPU and SoC performance, power, and thermal solutions. Led a team of 16 engineers, overseeing software architecture design and ensuring successful execution of deliverables Areas of expertise : GPU and SOC power and thermal management HW/FW/SW architecture GPU and SOC performance states and clocks/voltage management HW/FW/SW architecture Control Systems design and tuning - Close Loop Controllers, PID and heuristic models, DL based Perf and Power Controls RTOS, Embedded Systems, Computer Architecture, RISCV, ARM Details: 1. Team lead for Power and Performance software management coordinating the workflow of lower-level professionals while performing a set of similar or more complex duties for improving the power and performance states of NVIDIA's graphics and SoC processors 2. Interacting with external teams to collect ideas and use the same to scope new projects for Power and Performance Software management team 3. Mentoring multiple full time as well as intern engineers to deliver Power and Performance projects on NVidia’s graphics processors 4. Designing, implementing, and maintaining software for managing various power and performance states of NVidia’s graphics processors 5. Defining and implementing software modules as interfaces between user space tools and the NVIDIA's kernel device driver and developing test and benchmarking infrastructure for testing and validation of existing software.

Designed, implemented, and maintained software to manage power, thermal, and performance states of Nvidia’s graphics processors. Provided mentorship to junior engineers and contributed to innovation through filed patents. Areas of expertise : GPU power and thermal management architecture GPU performance states and clocks management architecture

Designed and implemented power and thermal control algorithms for the Pascal GPU family, developed software solutions for dynamic frequency/voltage scaling, and enhanced debugging tools to accelerate post-silicon bring-up. Led Perf, Power, Thermal GPU SW bring up validation. Position Duties: - Extended power capping algorithms to support a Pascal GPU family’s split logic and SRAM architecture, leading the design, implementation, validation, and bug resolution for product release. - Acquired expertise in software infrastructure solutions for dynamic frequency and voltage scaling, power, and thermal management, and implemented IO control interfaces from Kernel Mode Driver to User Mode. - Developed and enhanced tools for performance and power debugging by adding support for new interfaces, significantly accelerating post-silicon bring-up efforts, including tuning, characterization, and validation. - Ported Windows-based tools to Linux, ensuring compatibility with various Linux distributions to support cross-platform performance analysis and debugging. - Successfully led performance and power feature post silicon bring up for a Pascal GeForce GPU family.

Modified and optimized NVIDIA GeForce Power Management (Kernel) code by implementing a Class Hierarchical Structure, improving code organization, maintainability, and performance, while reducing binary size and enhancing consistency across the module.

Developed and tested various programs in C to support the Blue Cheetah App Ecosystem, ensuring functionality and performance through rigorous testing.

Provided technical support to USA clients, resolving network, OS, and software issues across servers, applications, VPN, antivirus, and more. Trained in Linux OS (RHCE), networking (CCNA), Windows Server, and computer hardware, ensuring effective problem-solving for both hardware and software-related challenges.

Electronics Engineering. Interested in Embedded Systems and Computer Architecture