RESPONSIBILITIES:

• Develop and integrate SOTA methods for efficient, large-scale training of ML models and support multi-platform deployment including automotive-grade edge compute devices
• Operate cross-functionally and serve a dual hat role to identify opportunities to improve production models, for example by improving model architectures and while trailblazing and generalizing involved methods and toolings to empower others
• Architect and develop tools for ML model evaluation and end-to-end validation to help ML engineers assess impact of their changes to downstream customers and modules
• Improve the utility of our vehicle data by building and improving the infrastructure for data sampling, data curation and data representation; improve versatility of our ML data format to support data reuse across models
• Operate cross-functionally and identify bottlenecks such as latency hot spots during training and deployment of ML models, while generalizing needed tools to empower others
• Scale our ML Ops architecture to the next level while taking advantage of heterogeneous clusters to maximize resource efficiency

QUALIFICATIONS:

• ML/ML Ops engineer with extensive experience in building large-scale data intensive distributed applications
• Experience in the full ML Ops cycle covering data cleansing, data sampling, data curation, training, testing, and deployment in the cloud and on edge compute platforms
• Prior experience of building and shipping ML models in a production environment
• Knowledge of SOTA methods in ML infrastructure domain and demonstrated curiosity and track record for keeping up with the literature
• Expert Python and PyTorch practitioner
• Familiarity with deployment and tuning ML models for edge devices
• Familiarity with C++