22 hours ago
Base Salary
$146k - $234k/yr
Responsibilities
- Design, implement, and train causal graph models, planning algorithms, and multi-objective optimization approaches.
- Build and maintain training and inference pipelines, experiment tracking, model versioning, dataset loaders, and reproducible ML infrastructure.
- Package trained models as containerized inference components conforming to specified APIs and deliver regular code revisions.
- Design synthetic-data generation and simulation campaigns that cover the model objective space.
- Present research results and author technical material for design reviews, workshops, status reports, and documentation.
- Coordinate with academic subcontractor researchers and collaborate with RF, DSP, and formal methods engineers.
Requirements
- 5+ years of experience and an MS or PhD in Computer Science, Electrical Engineering, Statistics, Applied Mathematics, or a related technical field.
- 2+ years of applied machine-learning research experience in causal inference, structural causal models, Bayesian networks, probabilistic graphical models, probabilistic programming, learned optimization, or planning.
- Strong Python software-engineering skills and production-quality PyTorch experience, including custom model implementation, training loops, and GPU performance profiling.
- Experience converting research prototypes into maintainable, tested, containerized code that runs unattended in uncontrolled environments.
- Working knowledge of multi-objective and constrained optimization and experience applying it to structured design spaces.
- Experience with experiment management, reproducible ML pipelines, and dataset versioning on a multi-person research team.
- Ability to implement methods from current research literature and explain model behavior and limitations to non-ML engineers and government stakeholders.
- US citizenship is required.
- Preferred qualifications include IARPA or DARPA research experience, communications or digital signal-processing knowledge, neuro-symbolic methods, graph-generation models, symbolic regression, executable-code generation, hardware-constrained inference, relevant publications, and willingness to obtain a Secret security clearance.
About Peraton
Peraton builds and integrates mission systems, cyber, space, and intelligence solutions for U.S. defense, intelligence, and civil agencies. It delivers enterprise IT, satellite and terrestrial communications, and spectrum management under government contracts. Founded in 2017, the company is headquartered in Reston, Virginia and is privately held by Veritas Capital.