Principal Architect, Performance Analysis and Modeling

📋 Description

• • d-Matrix Corporation is at the vanguard of technological transformation, dedicated to unlocking the immense potential of generative AI. We are a dynamic company pushing the boundaries of both software and hardware innovation, striving to redefine what's achievable in the AI landscape. Our core values are rooted in respect, collaboration, humility, and direct communication. We foster an inclusive environment where diverse perspectives are not only welcomed but are seen as the catalyst for developing superior solutions. We are actively seeking individuals who are passionate about confronting complex challenges and possess a strong drive for execution. If you're ready to discover your 'playground' and contribute to shaping the endless possibilities of AI, d-Matrix is the place for you.
• • This Principal Architect role is pivotal in accelerating AI application performance by operating at the critical intersection of hardware and software. You will play a key role in developing and optimizing solutions for emerging hardware technologies, including but not limited to DIMC, D2D, and 3D-DRAM, and for cutting-edge workloads such as generative inference. Our holistic acceleration philosophy spans across the entire system, encompassing efficient tensor cores, optimized storage and data movement strategies, and the co-design of dataflow and collective communication techniques.
• • As an integral member of our architecture team, your responsibilities will include in-depth analysis of the latest and most demanding ML workloads. This encompasses multi-modal Large Language Models (LLMs), Chain-of-Thought (CoT) reasoning models, and complex video and audio-generation tasks. You will be instrumental in contributing to both hardware and software features that will define the next generation of inference accelerators deployed in data centers.
• • Success in this role hinges on your ability to stay abreast of the latest advancements in ML architecture and algorithms. You will collaborate closely with a diverse range of partner teams, including Product Management, Hardware Design, Compiler Engineering, Inference Server teams, and Kernel Development. This cross-functional collaboration is essential for translating research insights into tangible product improvements.
• • Your day-to-day activities will be multifaceted and impactful. Primarily, you will analyze the intrinsic properties of emerging machine learning algorithms and workloads, meticulously identifying their functional requirements and performance implications. Concurrently, you will develop sophisticated analytical models designed to project performance metrics on both current and future generations of d-matrix hardware. Based on these analyses and projections, you will proactively propose novel hardware and software features that are crucial for enabling or significantly accelerating these advanced algorithms.
• • This role offers a unique opportunity to influence the future of AI hardware and software co-design. You will be at the forefront of innovation, working with state-of-the-art technologies and tackling some of the most challenging problems in the field. Your contributions will directly impact the performance, efficiency, and scalability of AI systems, making a tangible difference in how AI is deployed and utilized globally. The hybrid work model, with 3 days per week onsite at our Santa Clara, CA headquarters, offers a blend of collaborative in-office engagement and focused remote work, ensuring flexibility and productivity.