Blaize - Senior Staff Engineer - Artificial Intelligence/Machine Learning (8-12 yrs)

About the Role :

We are seeking a highly motivated and experienced Senior Staff Software Engineer to join our team and contribute to the development of our cutting-edge AI/ML software stack for the Blaize GSP (Graph Stream Processor).

You will play a crucial role in bringing up, optimizing, and debugging neural networks on our hardware, enhancing TVM features, and developing and maintaining the Model Conversion Tool.

This is a unique opportunity to work on a full-stack framework, directly impacting the performance and capabilities of our AI/ML platform.

Responsibilities :

- Neural Network Enablement : Bring up, test, and debug neural networks using popular ML frameworks like TensorFlow, PyTorch, and ONNX for the Blaize GSP. This includes working with pre-trained models, adapting them to our hardware, and ensuring optimal performance.

- TVM Enhancement : Bring up and enhance features within the TVM (Tensor Virtual Machine) framework to support the Blaize architecture and improve its functionality. Contribute to the ongoing development and maintenance of TVM integration.

- Performance Optimization : Analyze and enhance the efficiency and stability of neural networks running on the Blaize GSP. This involves profiling, identifying bottlenecks, and implementing optimizations to maximize performance.

- Model Conversion Tool Development : Develop and maintain the Model Conversion Tool software stack, which enables users to convert models from various frameworks to the Blaize format. This includes designing, implementing, and testing new features and ensuring compatibility with different model formats.

- Network Optimization : Work on network optimization techniques, including node fusion, graph lowering, and adding custom operations to improve the performance and efficiency of neural networks on the Blaize hardware.

- Performance Tuning : Profile and fine-tune the performance of neural networks to achieve optimal results on the Blaize GSP.

- Model Quantization : Implement and refine model quantization techniques to reduce model size and improve performance without significant accuracy loss.

- Software Development Best Practices : Adhere to software development best practices, including code reviews, version control, and testing.

Education and Experience :

- MS or BS in Computer Science or Electronics Engineering.

- 8-12 years of relevant experience or an equivalent combination of experience and education.

Required Knowledge, Skills, and Abilities :

- Programming Languages : Hands-on experience in C, C++, Python, NumPy, and OpenCL/assembly compute kernels.

- ML Frameworks : Hands-on experience with ML frameworks like TensorFlow, PyTorch, and ONNX.

- TVM Expertise : Strong knowledge of the TVM framework.

- Linear Algebra : Good understanding of linear algebra concepts.

- Network Optimization : Knowledge of network optimization, graph lowering, and fine-tuning techniques.

- Analytical Skills : Strong analytical and problem-solving skills.

- Software Design : Good understanding of algorithms, object-oriented programming (OOP) concepts, and software design patterns.

- Debugging Skills : Excellent debugging skills.

Added Advantages :

- Hardware Architecture : Experience or knowledge of hardware architecture is a plus.

- Full-Stack Framework Development : Experience with full-stack framework development like any of Multimedia frameworks, GStreamer, OpenVx, OpenMax, OpenGL, OpenGL-ES, Vulkan, Mesa, etc. is a plus.

- Driver Development : Experience in driver development on Linux platforms.

Mandatory Skills :

- C/C++

- Python

- NumPy

- OpenCL/Assembly

- TensorFlow

- PyTorch

- ONNX

- TVM