Role overview
We are looking for an Applied AI Engineer to connect our research with the reality of our partners’ environments that are constrained by hardware, power, and real world interference.
We work with customers operating in complex areas across robotics, simulation, aerospace and defense, manufacturing, logistics, industrial automation, and more. You will be responsible for taking our model, focusing on post-training, evaluation data, and integrations to ensure our customer's platforms work regardless of a messy, constrained tech stack and hardware.
You will embed with our partners to understand their actual problems, not just what they put in an RFP. You’ll look at their legacy control systems, latency challenges, and power needs and figure out how our AI helps them achieve something that was never possible before.
You'll be a key part of the feedback loop. When a model fails because of sensor noise or unexpected physics, you don't just log a bug. You figure out why, and you work across our team to fix the underlying architecture. You ensure we are building technology that survives contact with the real world, for years to come.
We're looking for a technical polyglot. You might have started in systems engineering, physics, or neuroscience and moved to ML, or the other way around. You know Python and PyTorch, but you aren't afraid of C++ or low-level hardware constraints.
Most importantly, you have high agency and want to be a part of an amazing team.
Responsibilities
- Embed with partners to solve the problems with our frontier AI/ML tools, informing our research and product development plan along the way…not just deploy software.
- Be the primary filter between the messy reality of the physical world and our research and technical staff, surfacing real commercial challenges and pain-points.
- Build and tune models, prototype, script, and patch (often in the field), turning ambiguous requirements into executable code.
- Build the connective tissue between our AI and the customer's reality, and then help them rethink the art of the possible. This means writing high-performance code (C++/Go/Python) that integrates our inference engine with legacy sensors, RTOS, and diverse hardware peripherals.
- Optimize complex ML models for survival in harsh computing environments.
- Leverage our simulation and world-model capabilities to validate operational plans before they touch physical hardware.
- Translate the probabilistic nature of AI into the deterministic language of industrial control systems and mission operators.
- Explain trade-offs to non-technical individuals and deep technical details to systems engineers, building the trust required to deploy autonomous systems in critical paths.
- 5+ years experience taking complex systems from prototype to production, within software engineering or applied AI/ML
- Strong experience in the ML stack (Python, Docker, Kubernetes, infrastructure-as-code, and CI/CD for ML pipelines) with competent systems programming skills (C++, Go, Rust, or Java), and ability to use modern AI coding tools
- Strong applied machine learning experience, specifically in the lifecycle of deploying, evaluating, and debugging models
- Experience in at least one of the following, with working knowledge of the others: Agents or policy learning (e.g., RL, planning, control theory, spatial reasoning) World models, simulation environments (Unity/Unreal, Omniverse, Isaac Sim), or model-based learning Perception, sensor fusion, or inverse dynamics models (IDMs)
- Agents or policy learning (e.g., RL, planning, control theory, spatial reasoning)
- World models, simulation environments (Unity/Unreal, Omniverse, Isaac Sim), or model-based learning
- Perception, sensor fusion, or inverse dynamics models (IDMs)
- Exposure to bridging the "hardware-software" gap: integrating AI inference with sensors, edge devices, RTOS, or legacy industrial networks
- Full-stack systems mindset: understanding of memory management, concurrency, networking, and APIs
- U.S. citizenship and ability to obtain and maintain a national security clearance (TS/SCI preferred)
- Ability to comply with export control requirements (ITAR/EAR)
- Experience, and comfort in, forward-type environments often found with partners across the industrial base, defense, intelligence, aerospace, and robotics environments at the edge
- Edge AI, inference optimization, or deployment in constrained settings (TensorRT, ONNX, or mobile inference as examples)
- Background in autonomous systems, control, or real-time systems Startup or early-stage engineering experience
- Understanding of secure systems engineering or DevSecOps experience in regulated industries, including degraded, intermittent, limited) networking constraints
- Open-source contributions or demonstrable applied systems work, or a portfolio of "side projects" that demonstrate AI/ML, and engineering curiosity
- ML & Research: Python, PyTorch, NumPy, OpenCV, Triton, CUDA for large-scale training, real-time inference, and applied CV/ML
- Pipelines & Experimentation: Kubeflow Pipelines and Airflow with continuous evaluation, A/B testing, and performance monitoring across training and production
- Backend & Systems: Java services with Redis and RabbitMQ, plus performance-critical C++ components; containerized with Docker and Kubernetes on GCP or on-prem
- Clients & Edge Software: Electron/React desktop apps, C# and C++ high-performance recorders, and mobile clients in Swift (iOS) and Kotlin (Android)
- Infra, Hardware & Other Deployments: Terraform-managed infrastructure, CI/CD via GitHub Actions and CircleCI; deployment to NVIDIA GPU clusters, air-gapped or on-prem environments, hardened Linux systems (FIPS/STIG), and constrained real-world hardware requiring model optimization, hardware-specific acceleration, and secure supply-chain practices
Benefits
- Competitive salary and meaningful equity
- Comprehensive health insurance including dental and vision insurance
- 401k