Senior AI/ML Engineer

SRC, Inc. is currently seeking a Senior AI/ML Engineer to lead the technical execution of machine learning and generative AI efforts supporting ISR, EW, advanced sensing, and decision science missions. This role owns problems, not tickets. You will take ambiguous mission needs, identify the technical approach, build the datasets and models that address it, produce defensible evidence that the result works, then guide it toward transition into a program. The work ranges across classical ML, deep learning, reinforcement learning, and the generative AI stack, and it demands judgment about which of those is the right tool for a given problem. You will lead tasks of moderate scope, guide the technical efforts of less experienced engineers, contribute to proposals and customer-facing technical material, and set the standard for experimental rigor on the team. The successful candidate is someone who has shipped models into use, has seen the gap between benchmark performance and field performance, and knows how to close it.
What You’ll Do
Lead the technical approach for AI/ML tasks of moderate scope, from problem formulation and data strategy through model development, evaluation, and transition
Design dataset strategies for domains where labeled data is scarce or expensive, including synthetic data generation, simulation-to-real transfer, active learning, and weak supervision
Own model development end to end: architecture and method selection, training, fine-tuning, distributed training where warranted, hyperparameter strategy, and error analysis
Build evaluation methodology using task-grounded metrics, statistically sound comparisons, ablations, adversarial and out-of-distribution test sets, and honest characterization of failure modes
Lead generative AI efforts including model selection across open-weight and commercial options, supervised and preference-based fine-tuning (SFT, DPO, GRPO), RAG architecture, agentic and tool-using systems, and guardrail design
Apply reinforcement learning where it fits the problem, and be able to articulate when it doesn't
Drive model efficiency and deployment readiness: quantization, distillation, pruning, export and runtime optimization, and the compute/latency/accuracy tradeoffs that govern fieldable systems
Establish and enforce reproducibility practices through experiment tracking, data and model versioning, containerized environments, and CI for ML code
Guide the technical and research efforts of less experienced engineers; review their work, develop their judgment, and raise the technical rigor of the team
Contribute to proposals, white papers, and internal R&D pitches, and assist in presenting findings and recommendations to customers
Partner with algorithm engineers, software engineers, and program staff to migrate models from research code into real-time and embedded applications
Identify technical risks early and propose alternative approaches before a program discovers them
What You’ll Bring
Bachelor's degree in Computer Science, Engineering, Mathematics, Physics, Statistics, or a related technical discipline, plus 4+ years of relevant experience; or a Master's degree plus 2+ years
Demonstrated record of taking ML systems from concept to working capability, not just prototype notebooks
Strong proficiency in Python and deep expertise in PyTorch (or equivalent), including custom training loops, datasets/dataloaders, mixed precision, and profiling
Practical experience with the modern LLM toolchain: Hugging Face Transformers, PEFT, TRL or equivalent, and serving/inference frameworks such as vLLM, TGI, or Ollama
Substantive experience with at least one of: generative AI and LLM fine-tuning, deep learning for perception or signal domains, or reinforcement learning
Rigorous approach to evaluation and the statistical literacy to back it up — confidence intervals, significance, and the difference between a real improvement and noise
Experience mentoring or technically leading other engineers
Ability to explain technical work and its limitations to program managers, customers, and executives
Excellent written and verbal communication
Ways to Stand Out
Experience with distributed training (FSDP, DeepSpeed, Ray) and large-scale data pipelines
Deployment of models to constrained or edge environments: ONNX, TensorRT, TorchScript, quantized inference, or embedded GPU/SoC targets
Multimodal model work including vision-language models, RF/sensor-conditioned models, or cross-modal retrieval
Experience applying ML to radar, SAR, RF, EO/IR, acoustic, or other sensor data, and familiarity with the phenomenology involved
Production MLOps experience: Kubeflow, MLflow model registry, feature stores, monitoring, and drift detection

Skills
Python, Machine Learning, Deep Learning, PyTorch, LLM, Generative AI