Staff Software Engineer, Physical AI Orchestration

Destro AI•New York, NY

About The Position

Destro AI is building the intelligence layer for robotics — the system that tells robots and humans what to do across complex real-world operations. Founded by former leaders from HAI Robotics, Dexterity, and Matic, Destro is built on a simple thesis: robotics has solved movement, but not decision-making. At the core of Destro is MothershipOS — a centralized brain that coordinates robots, humans, and workflows in real time. Unlike traditional systems that rely on local intelligence on each robot, Destro brings global intelligence across the entire operation. We recently closed our Seed round at one of the highest valuations for a robotics company at this stage, and are already working with large enterprise operators to deploy real systems in production environments. This is the most senior engineering hire we will make this year, and the person in it will shape how Destro builds software for the next several years. We got to production fast. What exists today is a set of services that were built separately, under deadline, to win pilots — and they work, but they were never designed to be one system. There is no unified architecture across them, no consistent versioning or release discipline, no automated way to push an update to a customer site, thin test coverage, and not enough telemetry to know a site is degrading before the customer calls. You are being hired to bring those services together into a single, coherent, production-grade platform that scales across sites — and to build the version control, release, and deployment discipline underneath it. Not through a rewrite; customers depend on this system every day. Through consolidation, hardening, and doing it in the right order. You will start hands-on in the codebase. As the team grows, you will grow into leading it. This role is ideal for someone who is equally comfortable: unifying disparate services into one architecture, building the CI, versioning, release, and observability layer from nothing, refactoring live code without breaking customers, working across engineering, deployment, and product to sequence the work, and leading and growing a software team. We are specifically looking for engineers who have built and shipped inside robotics companies. This is not a domain you can pick up on the job here — the failure modes of a robot fleet running live in a warehouse are not the failure modes of a web platform, and we need someone who already knows the difference.

Requirements

8+ years building software, with a substantial portion of it at robotics companies — AMR/AGV, warehouse automation, manipulation, autonomous vehicles, drones, or comparable fielded robotic systems
You have shipped software that ran on real robots, in real customer environments, at real scale — not in a lab or a research group
You have personally taken a system from pilot-grade to multi-site production, and can walk through what broke and how you fixed it
You have consolidated fragmented services into a unified platform, on a live system, without stopping delivery
You have introduced engineering process to a team that didn't have any, and gotten people to actually adopt it
Pragmatic about technical debt — you sequence fixes by risk, not by preference
Comfortable owning production, including on-call and live incidents
Able to operate independently in ambiguous, fast-moving startup environments
Strong communicator who can align engineers, deployment teams, and customers on the same technical plan
Wants to build and lead a team, but wants to stay close to the code
High ownership mindset with strong bias for execution
Deep Python; strong systems fundamentals.
ROS/ROS2, and direct experience with multi-vendor or heterogeneous robot fleets
Fleet management, teleoperation, or robot observability tooling — whether you used it or built it
Distributed systems, message-based architectures, and the failure modes of intermittently connected edge devices
Git at scale — branching models, monorepo or multi-repo tradeoffs, release management
Containerization, infrastructure-as-code, and modern CI/CD
OTA update systems and fleet device management
Linux/server environments, networking, cloud systems, and edge deployments

Nice To Haves

C++, Go, or Rust a plus
Prior integration work against WMS/WES platforms (Manhattan, Blue Yonder, Körber, SAP EWM) or with 3PL operators is a plus

Responsibilities

Bring today's separate services into one coherent platform with clear service boundaries, shared data contracts, and consistent interfaces
Design for scale: multi-site, multi-tenant, and multi-vendor robot fleets from day one
Eliminate duplicated logic, one-off scripts, and per-site special cases
Own the technical roadmap that takes us from the current architecture to the target one, incrementally and without downtime
Establish real version control discipline: branching strategy, code review standards, semantic versioning, release branches, and change tracking
Build a repeatable, automated path from commit to running on customer hardware
Implement over-the-air update push across cloud and on-site edge compute — staged rollouts, device provisioning, version pinning per site, and a rollback path that works at 4am
Guarantee we always know exactly what version is running where
Stand up CI and meaningful automated test coverage, including simulation and hardware-in-the-loop testing so changes can be validated without a robot on the floor
Build structured logging, metrics, tracing, and alerting across the deployed fleet
Harden the integration surface with customer WMS, WES, and ERP systems — retries, idempotency, schema versioning, graceful degradation when their systems go down
Establish on-call, incident response, runbooks, and postmortem practice
Own uptime, performance, and stability of the platform in production
Define and implement engineering best practices across the software organization — code review, documentation, testing standards, definition of done
Lead cross-functionally with robotics, computer vision, deployment, and product to sequence work and resolve technical tradeoffs
Mentor the engineers already here and raise the bar on every hire after you
Grow into leading the software team as we scale headcount over the next 12 months