Lead Agentic Systems Engineer

The Opportunity

MARESTRA is a high-profile, stealth-stage defense tech venture redefining the boundaries of maritime operational capabilities. Backed by top-tier Israeli and US Venture Capital firms, we are building an elite foundation at the ground floor.

This is a rare opportunity to join a promising startup at a very early stage in the emerging defense tech domain.

What We Aim to Achieve

MARESTRA is pioneering an innovative, platform-agnostic cognitive orchestration layer to deliver maritime security as a service, built upon networks of uncrewed autonomous surface vessels.

We are building a multi-tenant, cloud-native but edge-resilient "Mission Brain" that leverages advanced multi-agent orchestration frameworks, semantic guardrails, and sandboxed dynamic execution fields.

Our system translates complex operational logic and real-time parameters into deterministic, auditable software constraints at sea. By decoupling mission-critical cognitive intelligence from legacy hardware, we are establishing an entirely new paradigm of autonomous coordination, virtualized supervision, and forensic accountability.

Role Overview & Expectations

As our Lead Agentic Systems Engineer, you will architect the cognitive orchestration engine that powers our Mission Management System (MMS). Your focus will be on designing, deploying, and optimizing the multi-agent AI frameworks that enable a single human operator to seamlessly command, supervise, and orchestrate a fleet of heterogeneous Uncrewed Surface Vessels (USVs) simultaneously.

Your responsibility is to take complex maritime operational parameters, including vessel specifications, mission characteristics, tactical geofencing, and Rules for Use-of-Force protocols, and translate them into high-performance, real-time agentic workflows. This engine must dynamically manage situational awareness, prioritize mission-critical telemetry over narrow communication links, handle real-time troubleshooting, and empower the operator to rapidly react to evolving operational threats at sea. We are looking for a seasoned engineer who has moved past basic LLM experimentation and has concrete experience shipping highly deterministic, production-grade agentic architectures at scale.

Key Responsibilities

• Agent Framework Architecture : Design and optimize the core orchestration framework (utilizing production frameworks or custom implementations) that manages complex, multi-agent interactions, state management, and real-time concurrency across a fleet of uncrewed assets.
• Operational & Business Logic Implementation : Translate complex maritime product requirements (e.g., tactical geofencing, multi-USV search patterns, alert prioritization, and Rules for Use of Force protocols) into highly structured, context-aware tasks for AI agents to solve under operational constraints.
• Guardrails, Safety & Determinism : Develop advanced semantic guardrails, prompt engineering defensive layers, and validation pipelines to ensure agent reasoning is completely safe, reliable, and strictly conforms to Human-in-the-Loop (HITL) compliance rules before any payload execution.
• Real-Time Data Integration : Feed live backend APIs, maritime telemetry data, and hardware simulator streams into the agents' context windows and RAG pipelines, allowing the system to dynamically troubleshoot link failures and adapt to evolving threats.

Requirements

• Experience : 8+ years of experience in Backend Software Engineering, Systems Architecture, or Core AI Engineering, with expert-level Python and deep familiarity with asynchronous programming and distributed systems.
• Agentic Architectures : Proven track record of designing, deploying, and maintaining LLM-powered applications and Multi-Agent orchestration frameworks (e.g., CrewAI, AutoGen, LangChain, or custom variations) in production environments, as well as experience scaling state machines and handling state-concurrency or loop-prevention in multi-agent workflows.
• Core AI Design Patterns : Deep architectural understanding of advanced AI patterns, including multi-agent orchestration, complex RAG architectures, prompt injection mitigation, and reliable tool-calling/function-calling patterns.
• Operational Workflow Translation : Robust ability to break down non-technical, multi-platform operational flows, sensor data inputs, and system rules into precise logical tasks for autonomous agents to execute.
• Systems Thinking : Experience balancing high-stakes decision-support logic, strict policy guardrails (Human-in-the-Loop constraints), and technical real-world limitations like latency and intermittent link connectivity.

Preferred Qualifications (Advantage)

• Hands-on experience integrating real-time telemetry streams, defense-oriented middleware/protocols (e.g., ROS2, MAVLink, DDS, MQTT, or NMEA), or time-series data into agent contexts and RAG pipelines.
• Experience working with software-in-the-loop (SIL) simulators or mocking environments to evaluate multi-agent behavioral reliability before physical field deployments.
• Background or deep interest in defense tech, maritime robotics, autonomous command and control (C2) systems, or adjacent tactical safety-critical environments.
• Advanced degree (M.Sc./Ph.D.) in Computer Science, AI, Robotics, Data Science, Statistics, or a related highly quantitative field.