Artificial Intelligence Engineer

Artificial Intelligence Engineer at Harrison Clarke Artificial Intelligence Engineer at Harrison Clarke in Piedmont, California Posted in about 23 hours ago.

Type:

full-time

Job Description:

AI Engineer -

Defense Robotics Location :

San Francisco (On-Site)

Compensation :

$250,000 - $400,000 base + Equity Most autonomy work never leaves the simulator. The model performs, the demo lands, and then it meets a real world environment and falls apart the first time the lighting is wrong or something moves that wasn't supposed to. This role is the opposite of that. The systems here run in the field, in real world environments, with people depending on the output. An early-stage defense company is building the intelligence layer for fleets of robots: a foundation model that takes a natural-language command from a human operator and turns it into coordinated sensing, decision-making, and action across many machines at once. The hard, unsolved part is the model that makes them act like they share one mind. That is the part you would own. As an AI Engineer, you would architect across perception, reasoning, and control rather than picking up a slice someone else scoped. Day-to-day includes training and evaluating VLA models, building the architectures for multimodal fusion and continual learning, giving agents memory and tool use, and pushing all of it from research into real-time inference that runs onboard while the hardware is in motion. You would also be in the field for trials, watching your work hold or break under real constraints and fixing it on the spot. They're coming off a colossal funding round but are still early enough that you will help set the culture and the process. The work is demanding and the pace is fast but the payoff is that what you build gets deployed instead of archived.

The fit:

you have spent at least two years shipping AI models, ideally in robotics, agents, autonomous systems, or anything that has to work in real time. You are strong in computer vision, deep learning, or multimodal architectures such as VLMs and transformers. You understand distributed training, fine-tuning, and reinforcement learning, and you have personally carried something from prototype to product in a place that moved fast.

Non-negotiable:

the role requires access to U.S. export controlled information, so you must be a U.S. Person.

On offer:

meaningful equity, top-of-range salary package, fully covered medical, dental, and vision, strong PTO, relocation support where eligible, and a team pulled from the top of autonomy, AI, and national security. Backed by well-known defense and frontier investors. If the work you most want to do is autonomy that ships to real systems and matters when it gets there, this is worth a conversation.