Beschreibung

In this role, you will be at the forefront of developing cutting-edge tools for AI solutions, focusing on optimizing performance and integrating various model frameworks. Your work will involve everything from pipeline development to enhancing documentation and sharing insights with the team.

Anforderungen

• •BS/MS in Computer Science, Electrical Engineering or equivalent work experience
• •Proven hands-on experience with edge deployment frameworks such as OpenVINO or TensorRT
• •Familiarity with model optimisation techniques such as quantization, compression and pruning
• •Proficiency in AI application development using Python with extensive experience in ML libraries such as PyTorch and TensorFlow
• •Demonstrated ability to build and deploy end-to-end pipelines with quantized models
• •Experience in training models using transfer learning to leverage pre-trained models for new tasks
• •Experience with MLIR and ONNXRuntime
• •Strong problem-solving skills and ability to troubleshoot complex problems
• •Proficient Linux skills
• •Practical knowledge of agile development using tools such as Jira, Git and GitHub
• •Self-motivated with ability to thrive in a fast-paced collaborative startup environment

Aufgaben

• •Develop customer-facing components of the Voyager SDK
• •Compile high-level graphical representations of ML-based pipelines
• •Optimize memory usage and synchronization on heterogeneous hardware
• •Write low-level implementations of computer vision operators
• •Study research papers and open-source code for model deployment
• •Evaluate performance and accuracy of new models using the SDK
• •Implement image pre and post-processing operators
• •Collaborate with the compiler team to troubleshoot issues
• •Write decoders to transform inference tensors to metadata
• •Integrate industry-standard model frameworks within the SDK
• •Define metadata representations for common model types
• •Create libraries for evaluating model accuracy on Axelera AI hardware
• •Develop libraries for visually rendering inference results
• •Optimize end-to-end pipeline latency and throughput
• •Identify and resolve bottlenecks on supported hardware platforms
• •Develop profiling tools for customer pipeline analysis
• •Write and maintain comprehensive product documentation
• •Share knowledge with team members and promote best practices

Tools & Technologien

Benefits