Full-Time
Data Parallel Accelerator Post-Silicon Performance Lead
Posted on 7/1/2025
Rivos
201-500 employees
Custom RISC-V server hardware for enterprises
No salary listed
Austin, TX, USA + 1 more
More locations: Santa Clara, CA, USA
Hybrid
 Hardware Engineering (2) |
|---|
| , |
- Deep expertise in General-Purpose Graphics Processing Unit architecture and microarchitecture
- Strong programming skills in C/C++
- Strong programming skills in Python
- Solid understanding of machine learning and deep learning workloads and benchmarks
- Experience with performance counters and profiling techniques
- Knowledge of performance improvement concepts including bottleneck analysis, latency hiding, speculative execution, resource scheduling, buffer sizing, and replacement policies
- Experience with embedded systems (bare-metal testing/debugging is a plus)
- Excellent teamwork, ownership, and communication skills; ability to thrive under aggressive schedules and adapt quickly
- Bachelor’s degree with 12+ years of experience in a relevant field
- Master’s degree with 10+ years of experience in a relevant field
- PhD with 5+ years of experience in a relevant field
- Lead cross-functional performance validation: Analyze workloads and microbenchmarks in emulation and post-silicon environments, ensuring strong correlation with cycle-accurate models and RTL
- System-level performance optimization: Measure and tune workloads (Generative AI, data analytics) for optimal performance per watt
- Collaborate across teams: Work closely with design, architecture, systems, and software groups to enable enterprise use-case performance measurements
- Power and performance correlation: Integrate silicon power measurements with simulation and full-chip projections to drive hardware/software tuning
- Performance infrastructure automation: Develop and automate tools for performance measurement, debug, and reporting
- Debug and tuning: Conduct system-level power and performance debugging, including silicon register tuning to meet aggressive performance targets
- Drive innovation: Influence architectural decisions and validation methodologies to ensure our platforms remain at the forefront of the industry
- Experience optimizing ML/DL workloads and LLMs at the system level
- Familiarity with SIMT processing, cache, and memory hierarchies
- Hands-on experience with embedded systems (bare-metal testing/debugging is a plus)
- Knowledge of power and area analysis and correlation with simulation
- Experience with post-silicon validation environments and emulation platforms
Rivos designs and manufactures custom high-performance server hardware based on the RISC-V open ISA for data centers and enterprise IT. Its products are purpose-built to deliver strong compute performance, energy efficiency, and security for demanding workloads, through hardware tailored to specific tasks rather than off-the-shelf designs. The company differentiates itself by leveraging the flexibility and cost advantages of RISC-V to create bespoke servers that meet enterprise needs, offering close collaboration and support to ensure seamless integration in large-scale infrastructure. Its goal is to serve cloud providers, data centers, and other large IT operators with tailored, dependable server solutions that optimize performance and total cost of ownership.
Company Size
201-500
Company Stage
Series A
Total Funding
$250M
Headquarters
Santa Clara, California
Founded
2021
Simplify's Take
What believers are saying
- Meta's $2B acquisition in September 2025 validates RISC-V for AI chips.
- Rivos raised $250M in April 2024 to scale AI inference GPUs.
- Canonical partnership enables Ubuntu on Rivos RISC-V data centers.
What critics are saying
- Meta redirects Rivos roadmap to internal AI by 2026, halting enterprise sales.
- Nvidia's 20-30% H100 price cuts in 2025 lock in CUDA ecosystems.
- RISC-V fragmentation demands proprietary optimizations, blocking portability.
What makes Rivos unique
- Rivos builds high-performance RISC-V SoCs with proprietary GPGPU accelerators.
- Rivos delivers CUDA-compatible software stacks for seamless AI transitions.
- Rivos taped out 3.1 GHz RISC-V processor for data center efficiency.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Flexible Work Hours
Growth & Insights and Company News
6 month growth
↑ 50%1 year growth
↑ 53%2 year growth
↑ 51%Terms of the deal have not been disclosed
Meta Platforms (NASDAQ: META) has acquired chip startup Rivos to enhance its AI capabilities and reduce reliance on external suppliers. This move, announced on September 30, 2025, aims to integrate Rivos's RISC-V-based processors and GPUs, accelerating Meta's custom AI chip development. The acquisition reflects a broader trend of tech giants pursuing vertical integration in AI hardware, potentially impacting companies like Nvidia. Rivos was previously seeking funding at a $2 billion valuation.
Meta is acquiring chip startup Rivos Inc. for around $2 billion to enhance its AI hardware capabilities and reduce reliance on Nvidia GPUs. Rivos, known for its RISC-V–based AI GPU, raised $250 million in April 2024. This acquisition aligns with Meta's strategy to develop in-house AI chips by 2026, aiming to cut AI compute costs. The move is part of a broader trend among tech giants to control AI chip development, challenging Nvidia's market dominance.
U.S.-based semiconductor startup Rivos is aiming to raise between $400 million and $500 million to develop advanced graphics processing units (GPUs) designed to compete directly with NVIDIA in the fast-growing artificial intelligence (AI) chip market.
Santa Clara, California-based Rivos is reportedly developing a chip optimized for artificial intelligence inference.