Senior Quantization Framework Development Engineer (ML)

at  Axelera AI

ABOUT US

Axelera AI stands at the forefront of the semiconductor industry, driving advancements in artificial intelligence acceleration.
In just over 2 years, we’ve assembled a team of over 140 brilliant engineers, developers, and business experts from across the globe. Guided by the brightest minds in the AI sphere and supported by renowned innovation leaders like IMEC, we’ve secured a total of US$50 million in funding.
Our unwavering commitment to innovation has firmly established us as a global industry pioneer.

POSITION OVERVIEW

Deep Neural Networks are powerful tools to enhance edge intelligence, including computer vision, drones, and robotics. However, edge devices are resource constraint, requiring model optimization for efficient algorithm deployment. One optimization strategy, quantization, reduces neural network bitwidths. We are expanding our team with a Senior ML Engineer to expand and refine our quantization framework within our Voyager SDK, enabling efficient DNN deployment on our advanced AI accelerator.

REQUIREMENTS:

• Python Proficiency:

Possess a minimum of 5 years of hands-on experience with Python, including proficiency in libraries such as Numpy and Scipy.

• Deep Learning Frameworks:

Demonstrate at least 4 years of practical experience with PyTorch, with a preference for proficiency in Torch FX, ONNX, and Tensorflow.

• Neural Network Understanding:

Showcase a strong understanding of the inner workings of recent Convolutional Neural Networks (CNNs), Multi-Layer Perceptrons (MLPs), Recurrent Neural Networks (RNNs), and transformers.

• Coding Excellence:

Proventrack record of writing high-quality, efficient, and maintainable code, with an emphasis on best practices and coding standards.

• Problem-Solving Skills:

Exhibit a problem-solving mindset, with the ability to diagnose, debug, and resolve complex software issues efficiently.

• Communication Skills:

Possess excellent oral and written communication skills in English, with the ability to convey technical concepts to diverse audiences.

• Team Collaboration and Autonomy:

Be a collaborative team player, capable of working effectively within a team, while also demonstrating the ability to autonomously plan and execute tasks.

• Sense of Responsibility and Ambition:

Demonstrate a strong sense of responsibility, commitment, and a drive to achieve ambitious goals in a dynamic and fast-paced environment.
- Experience implementing a deep learning quantization algorithm from scratch & contributing to a deep learning framework - highly preferred.

KEY RESPONSIBILITIES:

• Lead Exploration and Implementation of Quantization Methods:
  Spearhead the study and assessment of existing and emerging quantization methods for neural networks.

Drive the exploration and implementation of innovative quantization techniques to optimize model performance on resource-constrained edge devices.

• Framework Development and Maintenance:

Take a hands-on approach in developing, enhancing, and maintaining the quantization framework within the Voyager SDK.

Conduct rigorous testing and documentation to ensure the robustness, reliability, and scalability of the framework.

• Collaboration with Software Developers:

Collaborate closely with software developers to integrate and refine algorithms, fostering continuous improvement in algorithm optimization.

Contribute to the enhancement of overall SDK quality through collaborative development efforts.

• Results Communication and Code Reviews:

Effectively communicate research findings, insights, and results to both technical and non-technical stakeholders.

Actively participate in code reviews, providing constructive feedback and ensuring code quality across the team.

• Stay Abreast of Industry Trends:

Stay informed about the latest advancements and trends in machine learning, deep learning, and neural network optimization.
Act as a knowledge resource within the team by sharing insights on industry best practices and emerging technologies.