Swarm Engineer - Multi-Agent Task Planning

Who is Recruiting from Scratch: Recruiting from Scratch is a specialized talent firm dedicated to helping companies build exceptional teams. We partner closely with our clients to deeply understand their needs, then connect them with top-tier candidates who are not only highly skilled but also the right fit for the company’s culture and vision. Our mission is simple: place the best people in the right roles to drive long-term success for both clients and candidates.

Location: Phoenix, AZ

Company Stage of Funding: Early-Stage / Defense Technology Startup

Office Type: Onsite, Full-Time

Salary: Competitive (based on experience)

Security / Eligibility: Must be eligible to work on export-controlled projects

We’re representing a defense-focused robotics company building low-cost, autonomous swarm systems designed to operate in complex, high-stakes environments. Their core product is a fleet of autonomous unmanned ground vehicles (UGVs) that operate independently or in coordinated swarms to execute advanced, multi-domain missions.

The company is led by founders with decades of experience across autonomous vehicles, robotics, aerospace, and national security. Their mission is to deploy intelligent, attritable swarm systems that solve critical defense challenges while maintaining scalability and cost efficiency.

As a Swarm Engineer – Multi-Agent Task Planning
, you will design and deploy multi-modal action models that enable real-time coordinated swarm behaviors. This is not a perception role — the focus is on decision-making and action selection at both individual vehicle and swarm levels.

• Architect, train, and iterate on multi-modal action models that select tactical macro-actions based on rich contextual inputs.
• Design model architectures that fuse heterogeneous data sources (local perception outputs, swarm state, mission objectives) into unified decision representations.
• Develop and apply reinforcement learning approaches (online and offline), including transformer-based sequence modeling for swarm coordination.
• Optimize models for real-time edge execution using quantization, distillation, and efficient architecture design.
• Build and maintain full ML pipelines from data collection and curation through training, evaluation, and field deployment.
• Integrate action models into broader autonomy stacks alongside navigation and planning systems.
• Deploy and validate trained policies on physical UGV swarms in field environments.
• Write robust production-quality Python and C++ code.

This role operates at the core of swarm intelligence — translating situational awareness into coordinated, tactical action.

The ideal candidate is a strong ML engineer with deep expertise in action-oriented model design and multi-agent coordination systems.

• Strong mathematical foundation in neural networks, transformers, reinforcement learning, and statistics.
• Proficiency in Python and C++.
• Experience with PyTorch or Tensor Flow.
• Experience training and deploying models that generate actions or macro-actions (e.g., reinforcement learning, planning-as-inference, VLA models).
• Familiarity with multi-agent coordination concepts such as task allocation, distributed decision-making, or swarm behaviors.
• Experience optimizing and deploying ML models on resource‑constrained or edge hardware.
• Eligible to work on export‑controlled projects and able to relocate to Phoenix, AZ.

• Experience with policy gradient methods (e.g., PPO).
• Experience with multi‑agent task planning algorithms (auction‑based allocation, distributed scheduling, swarm coordination).
• Familiarity with ONNX, Tensor
  
  RT, and edge deployment tool chains.
• Prior experience in robotics, autonomous vehicles, or unmanned systems.
• Experience with simulation environments and synthetic data generation for training multi‑agent policies.
• Experience owning an end‑to‑end data‑to‑production model pipeline.
• Academic publications in related fields (NeurIPS, AAAI, IROS, ICRA, JAIR, etc.).

• Compensation: Competitive salary based on experience.
• Work Model: Full‑time, onsite in Phoenix, AZ.
• Impact: Direct ownership of swarm‑level intelligence systems in a defense robotics platform.
• Growth: Opportunity to define and scale multi‑agent action architectures from the ground up.

Mission Alignment: This role is ideal for engineers motivated by applying machine learning to real‑world, high‑impact autonomous systems in defense contexts.

Equal Opportunity: The company is committed to equal employment opportunity and complies with all applicable federal, state, and local employment laws.

Salary Range: $150,000-$160,000 base.