Dieses Stellenangebot ist nicht mehr verfügbar
Über
- Builds software pipelines that adapt neural networks (such as Hugging Face, Ultralytics, or custom models) for deployment on Mythic's hardware.
- Develops advanced quantization-aware and analog-aware retraining algorithms leveraging PyTorch and ONNX.
- Hardens networks to analog effects via advanced network regularization.
- Models analog effects and their impact on network performance.
- Works cross-functionally to validate and debug hardware.
- Contributes to the co-design of next-generation hardware.
- Brings up and customizes neural networks.
- Optimize Mythic's analog-aware software toolchain for network accuracy, latency, and ease-of-use.
- Design algorithms and tools for Mythic's neural network conversion pipeline.
- Build high-fidelity, computationally-efficient hardware models.
- Contribute to silicon bring-up, debugging, and validation.
- Improve software through refactoring, testing, documentation, and other engineering best practices.
- Stay current with advances in deep learning research and neural network frameworks.
- Bachelor's degree in Computer Science, Mathematics, or a related field.
- 5+ years of software experience in a production environment.
- Experience working on complex problems with algorithm-heavy code.
- Commitment to quality and engineering excellence.
- Strong communication skills.
- MS/PhD in Computer Science, Mathematics, or related field.
- Hands-on experience with modern neural network frameworks.
- Familiarity with state-of-the-art neural network architectures.
- Experience training neural networks with hardware-aware techniques, including quantization, pruning, or model-size limitations.
- Experience with MLOps practices, including model versioning, CI/CD pipelines for ML, model deployment, and monitoring.
- Experience owning critical APIs with a large user base.
- Contributions to open-source software.
Sprachkenntnisse
- English
Hinweis für Nutzer
Dieses Stellenangebot wurde von einem unserer Partner veröffentlicht. Sie können das Originalangebot einsehen hier.