Lead Vector Compute Architect at Bolt Graphics

Sunnyvale, California, United States -

Full Time

Start Date

Immediate

Expiry Date

25 Aug, 26

Salary

220000.0

Posted On

27 May, 26

Experience

5 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Vector Compute Architecture, Data Parallel Microarchitecture, Performance Modeling, RTL Development, C/C++, SystemC, Memory Hierarchy, Cache Coherency, Workload Analysis, SoC Systems, Hardware-Software Co-design, GPU Architecture

Industry

Semiconductor Manufacturing

Description

About the role: We are looking for an experienced and highly motivated Lead Vector Compute Architect to lead the architecture definition and technical direction for Bolt’s next-generation GPUs. The ideal candidate will have strong expertise in data parallel compute unit architecture development, performance modeling, data path integration, and cross-functional collaboration across hardware, software, and systems teams. This role involves defining scalable and high-performance architectures for advanced compute workloads including graphics, HPC, and system management. This role is on-site and requires someone to be local to the Bay Area. What you'll do: Define data parallel microarchitecture satisfying ISA constraints. Drive architecture tradeoff analysis for performance, power, area, bandwidth, latency, and scalability. Develop and review system architecture specifications, interface definitions, and microarchitecture requirements. Collaborate with RTL, verification, physical design, firmware, software, and system teams throughout the development cycle. Lead performance modeling, workload analysis, and bottleneck identification using C/C++/SystemC or similar modeling environments. Define memory hierarchy, coherency architecture, and cache structures. Work closely with verification teams to define architectural test plans and validation strategies. Support silicon bring-up, debug, performance tuning, and post-silicon optimization. Contribute to long-term technology and product roadmap planning. Qualifications: Strong understanding of modern data parallel microarchitectures and subsystem integration. Bachelor’s or Master’s degree in Electrical Engineering, Computer Engineering, Computer Science, or related field. 6+ years of experience in modern data parallel microarchitecture including: Workload characterization and profiling Performance modeling Out-of-order data dependency and control Utilization / occupancy optimization High-performance architecture design techniques Experience with one or more of the following: CPU/GPU/NPU architectures NoC/interconnect architectures Cache coherency protocols (CHI/ACE/CXL) High-speed interfaces (PCIe, UCIe, Ethernet) Memory systems (DDR, LPDDR, HBM, GDDR) Power, performance, and area optimization Strong knowledge of RTL development and verification methodologies. Experience with architecture modeling and performance analysis tools. Familiarity with firmware/software interaction in complex SoC systems. Excellent problem-solving, communication, and leadership skills. Compensation Range: $180,000–$220,000 per year (California). This range represents the anticipated base pay for this role; the final offer may vary based on qualifications, experience, and location. Benefits: Medical, Dental, & Vision - 100% covered premiums Equity - Stock Options 401(k) match WFH Hardware Bolt is committed to building a diverse and inclusive environment in which we recognize and value each other’s differences as well as fostering a culture that promotes its core values: Professionalism, Integrity, and Respect. As an equal opportunity employer, all qualified applicants will receive consideration for employment without regard to race, color, religion, sex, sexual orientation, gender identity, genetic information, national origin, age, disability, or status as a protected veteran.

Responsibilities

Lead the architecture definition and technical direction for next-generation GPUs, focusing on data parallel compute units. This includes driving tradeoff analysis for performance, power, and area while collaborating across hardware and software teams.