Forterra is seeking to solve the problem of building advanced models for understanding complex agent interactions in dynamic environments, with a focus on ground autonomy solutions for the U.S. Department of Defense.
Requirements
- Deep understanding of probabilistic generative models for sequential forecasting tasks
- Strong experience with Transformers and graph-based modeling for complex relational data
- Familiarity with multimodal sensor processing and representation learning
- Proven ability to work with uncertainty quantification and latent variable inference
- Experience implementing models in PyTorch/TensorFlow and deploying them in production environments
- Strong foundation in applied deep learning, especially variational inference and attention mechanisms
- Solid software engineering background with emphasis on clean, modular, and optimized implementations
Responsibilities
- Develop neural network architectures for modeling sequential agent interactions using generative modeling approaches
- Implement Transformer-based models that encode agent and contextual features with attention mechanisms for capturing interactions
- Research and integrate goal-conditioned models that leverage structured representations before generating complete sequences
- Prototype and optimize learning models into lightweight modules suitable for real-time inference
- Collaborate with cross-functional teams to ensure seamless integration of models into the broader AI stack
- Benchmark and visualize model performance using real-world and simulated datasets
- Optimize models for low-latency inference and interpretability of output distributions
Other
- M.S. or Ph.D. in Computer Science, Machine Learning, Robotics, or related field
- 3+ years of experience with sequential modeling, generative architectures, or forecasting systems
- CLEARANCE ELIGIBILITY - This position may require eligibility to obtain and maintain a U.S. security clearance.
- Travel requirements not specified
- Must be willing to work in the Palo Alto office
