Perception Engineer

Mach Industries • San Francisco, CA

3d • $125,000 - $220,000

About The Position

Mach Industries is a defense technology company founded in 2023 that develops autonomous defense platforms. The company has approximately 350 employees and focuses on scalable, decentralized defense systems. As a Perception Engineer, you will help build an AI-forward autonomy stack for contested environments where GPS and other sensing may be unavailable or unreliable. You will design, train, and deploy vision and multi-sensor perception systems for navigation, targeting, and automatic target recognition across Mach’s product lines, bringing research-grade algorithms into real-world deployments.

The role spans deep learning, computer vision, and embedded systems. You will develop perception algorithms for EO/IR and multi-spectral imagery, build datasets and model-training and evaluation workflows, optimize models for real-time embedded inference, and combine vision with auxiliary sensor inputs. You will also create tools for analyzing simulation, hardware-in-the-loop, and flight data; test algorithms with high-fidelity sensor models and real-world datasets; and monitor system health, drift, and graceful degradation. The position requires production C++ and Python experience, at least seven years of relevant experience, and a track record of shipping machine-learning models to production.

Compensation is listed at $125,000–$160,000 USD for Engineer and $165,000–$220,000 USD for Senior Engineer. The company describes performance-based incentives, equity participation, healthcare, dental and vision plans, retirement savings, paid time off, and professional-development funding as elements of its compensation and benefits offering. Access to certain controlled information may require U.S. citizenship or another qualifying status, or eligibility to obtain the required authorizations; an offer may be conditioned on authorization without export-license sponsorship.

Requirements

• Production C++ on Linux and Python for machine-learning and tooling, including profiling, optimization, and rigorous testing.
• Experience working with CUDA backends for performance optimization and debugging.
• Strong experience with modern detection, segmentation, and tracking methods, and training or fine-tuning models in PyTorch.
• Experience building large, diverse datasets; labeling and quality-assurance pipelines; augmentation; experiment tracking; and reproducible training.
• Hands-on experience with model compression such as INT8/FP16, runtime optimization, and real-time constraints.
• Experience with EO/IR imagery and real flight or test data in challenging environments.
• At least 7 years of experience, with a BS, MS, PhD in CS/EE/Robotics or a similar field, or equivalent experience.
• A track record of shipping machine-learning models to production.
• For access to information subject to U.S. export regulations, the applicant must meet a listed qualifying status or be eligible to obtain required authorizations; employment may be conditioned on authorization without sponsorship for an export license.

Nice To Haves

• Multi-modal perception experience combining EO/IR with radar, LiDAR, or RF at the decision or feature level.
• Experience with robustness and safety practices, including adversarial or rare-event testing, long-horizon reliability metrics, and dataset-shift or drift monitoring.
• Experience with physics-aware imaging, including radiometric correction, NUC/FFC, atmospheric-effects modeling, and synthetic data or simulation for coverage.
• Experience with MLOps and data infrastructure, including SQL/Parquet, dataset/versioning tools, CI-based validation, and scalable multi-GPU training.

Responsibilities

• Build and refine detection, segmentation, and tracking architectures for EO/IR and multi-spectral imagery, including datasets, training recipes, and generalization to long-tail and degraded conditions.
• Establish training and evaluation pipelines using metrics such as PR/ROC, mAP, latency, and robustness suites; implement continuous regression testing and model-update loops based on field data.
• Optimize models for real-time embedded inference through quantization, pruning, and runtime optimization; profile CPU/GPU performance and meet throughput and latency targets on Jetson-class hardware.
• Combine vision outputs with auxiliary sensing such as radar, LiDAR, or RF cues for confirmation, denial, association, and track management using decision-level fusion.
• Create visualization, triage, and root-cause tools for simulation, hardware-in-the-loop, and flight logs; define end-to-end test plans with hardware and flight teams.
• Instrument health metrics and drift detection, support graceful degradation, and write tests and documentation mapped to performance requirements.
• Perform simulation-based testing with high-fidelity sensor models and validate algorithms using real-world datasets.

Benefits

• Healthcare, dental, and vision plans.
• Retirement savings plans.
• Paid time off.
• Professional-development funding for continuing education, training, and career growth.
• Performance-based incentives tied to individual and company performance.
• Stock options or equity participation.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior