Key Responsibilities

• Develop and implement quantization of existing AI models for various fields (e.g., LLM, Vit and Dit) for our unique optical AI accelerator, enabling inference with low-bit weights and activations without compromising model accuracy.
• Collaborate with system engineers to deploy, analyze and enhance the performance of quantized models on our optical processor.
• Stay informed about the latest developments in model quantization, analog computing, and AI acceleration. Utilize this information to aid and inform hardware design.
• Work closely with cross-disciplinary teams, including software developers, data scientists, and product managers, to integrate optimized AI models into comprehensive solutions.

Qualification & Skills

• 3+ years of industry and research experience in machine learning algorithms, with a focus on model quantization techniques
• Master’s or Ph.D. in Computer Science, Electrical Engineering, or a related field with a focus on machine learning
• Proficiency in Python and machine learning frameworks such as TensorFlow or PyTorch
• In-depth understanding of numerical representations used in machine learning and quantization techniques
• Experience with tools for quantization and model compression (e.g., ONNX, TVM)
• Excellent problem-solving skills and ability to work collaboratively across teams
• Strong communication skills for conveying technical insights and hardware requirements

Preferred Qualifications

• Experience with model quantization for LLMs
• Experience with model quantization for analog computing.
• Publications or patents in AI model quantization or hardware-aware AI optimization.