Dawei Huang

About Dawei Huang

Seasoned engineering executive and a founding member of the team at SambaNova, now leading the development of our inference cloud platform—the foundation for large-scale generative AI solutions. I guide the teams responsible for maximizing the performance, efficiency, and scalability of large language models (LLMs) in production.My journey here began with the initial concept, taking the system from PowerPoint to a functional product. This included early AI workload analysis, pre-silicon software bring-up, and the architecture of our scalable systems, culminating in the productization of our current SN40L platform and the launch of our SambaCloud services.This end-to-end experience—from founding engineer to VP—provides a deep, full-stack perspective that is critical for optimizing the entire AI pipeline and is a key differentiator in delivering state-of-the-art, production-ready AI infrastructure to the enterprise.My focus is on solving the industry\'s most challenging problems in AI inference, mentoring world-class engineering teams, and transforming cutting-edge research into robust cloud services.

Experience

1. Vice President of Engineering - Inference Cloud & Llm Performance

   SambaNova

   Nov 2017 — Present · CA, US

   Leading the engineering organization for SambaNova\'s flagship inference cloud service. My mission is to build the most powerful and efficient platform for deploying and serving large generative AI models.Organizational & Technical Leadership — Building, mentoring, and scaling high-performing engineering teams in ML systems, cloud services, and performance optimization.Platform Vision & Strategy — Setting the technical architecture and roadmap for our inference cloud, ensuring alignment with business objectives and superior customer outcomes.Full-Stack Performance Optimization — Driving initiatives to achieve best-in-class inference latency, throughput, and cost-efficiency across the entire technology stack.Cross-Functional Execution — Bridging research, hardware, software, and product teams to deliver robust and innovative AI cloud solutions.Previous Roles at SambaNova: Founding Engineer & Senior Director of EngineeringPlayed a leadership role in the architecture, design and productization of the SambaNova RDU and SN40L system from initial concept through production, guiding the team\'s work on early workload analysis, system design, and system bring-up.Contributed to scaling the engineering organization and the delivery of key strategic initiatives, including Samba-1 CoE and SambaCloud.Led the engineering team responsible for delivering world-record inference performance, including:→ The first to achieve tokens/sec for Llama3 8B→ World\'s fastest inference for Llama 405B and DeepSeek-R1 at launch time

Education

• UC San Diego

  Master's degree, Electrical and Computer Engineering

• Tsinghua University

  Bachelor's degree, Precision Instrument

Skills

• Debugging
• System Architecture
• Cmos
• Integrated Circuits (Ic)
• Semiconductors
• Ic
• Microprocessors
• Machine Learning
• Simulations
• Signal Integrity
• Verilog
• R&D
• Hardware
• Serdes
• Data Analysis
• Hardware Architecture
• Deep Learning
• Asic