Physical AI needs a production system

The Role

You will be one of the earliest engineers shaping both the technical architecture and the product direction of LatentWorlds. This is a deeply hands-on role for someone who can design systems, ship hard parts, work with users, and help decide what platform capabilities should exist.

The scope is broader than a typical senior engineering role. You will help decide not only how we build, but what we build, the quality bar we hold, how we work, and how the engineering team grows. We are looking for someone with 10+ years of experience, ideally including leadership and ownership of products or major product areas.

The role is intentionally broad, spanning distributed systems, product engineering, infrastructure, ML workflows, and customer work. You should be excited by infrastructure that can grow from low-level primitives into higher-level products, while staying grounded in what users actually need. We also expect engineers to use AI assistants well and improve the systems around AI-assisted engineering.

What You Will Work On

• Core systems and product primitives: build durable backend and platform foundations.
• Full-stack execution: move across backend, frontend, APIs, tooling, and workflows.
• Data and ML workflows: shape the path from robot data to debugging, evaluation, training, and iteration.
• Infrastructure and operations: improve observability, reliability, deployment, and developer workflows.
• Customer, product, and team work: partner with users, turn messy needs into product decisions, and help raise standards, hiring, culture, and AI-assisted workflows.

Who We Are Looking For

We care most about talent, ownership, judgment, and learning speed. Strong candidates usually show:

• 10+ years building production systems with real ownership.
• Leadership across teams, engineers, products, or major product areas.
• Broad range across backend, product, infrastructure, and adjacent areas.
• Strong systems fundamentals and product judgment.
• Execution under ambiguity, including direct customer work when needed.
• Clear communication, hands-on craft, and effective use of AI assistants. Rust is a plus.

The Role

You will build distributed systems that sit on the critical path of robot deployments and underpin higher-level platform capabilities. The hard part is not isolated backend plumbing; it is designing platform primitives that stay dependable under real constraints: unstable networks, large payloads, time alignment across sensors, user-defined workflows, and customers who need clear guarantees.

We are early and the scope is intentionally flexible. We want someone who enjoys owning ambiguous problems and turning them into clean primitives that other platform layers can build on.

What You Will Work On

• Ingestion reliability: resumable uploads, backpressure, durable acknowledgements, and end-to-end integrity checks.
• Storage, indexing, and retrieval: manifests, metadata models, and APIs that keep time-aligned access cheap and fast.
• Workflow and execution primitives: dataset versions, provenance hooks, execution substrates, and new services that higher-level pipelines and applications can build on.
• Ops and governance: identity, RBAC, audit trails, retention and deletion controls, and the observability metrics that keep us honest.
• Platform expansion: define and build new services, APIs, and product-facing capabilities where the platform needs to grow.

Who We Are Looking For

We care most about talent, ownership, and learning speed. Strong candidates usually show:

• Production experience shipping distributed systems that other engineers relied on.
• Systems thinking around outages, retries, backpressure, consistency, and correctness at the edges.
• API and data-model craft that stays stable under real product pressure.
• Clear communication and strong programming fundamentals.
• Working Rust knowledge, especially async Rust.

Bonus points for Rust async or systems work, multi-tenant SaaS infrastructure, protocol or performance work, and exposure to robotics data formats or time sync.

The Role

You will lead hands-on robotics and ML work that makes the platform roadmap real. This is not a paper-only research role. The output we care about is a loop that runs: collect, curate, train, evaluate, repeat.

Because we are early, you will have unusual leverage. The same person can shape how data is captured on the robot, represented in the platform, and turned into datasets, evaluations, and model-facing workflows that users can build on.

What You Will Work On

• Data collection on real robots: task design, instrumentation, time sync, and capture policies that survive messy reality.
• Dataset construction and QC: define episodes, metadata, versioning, and automatic checks for dropped topics, timestamp gaps, and corrupted payloads.
• Training and evaluation: train baseline models, build evaluations that catch regressions, and document what changed between runs.
• Tight platform feedback: work with the systems team so retrieval, provenance, and exports match how training actually works.
• Loop quality: help create an end-to-end path from robot recording to reproducible datasets, baseline models, and faster iteration.

Who We Are Looking For

We care most about talent, ownership, and learning speed. Strong candidates usually show:

• Depth in robotics or ML, ideally at PhD level or equivalent in practice.
• Hands-on work with real robots and real sensor data, not only simulation.
• Experience training models from multi-modal data and diagnosing why they fail.
• Ability to write production-quality code for experiments and tooling. Python is fine; Rust/C++ is a plus.
• Clear communication about experiments, assumptions, and results.

Bonus points for manipulation learning, data-centric ML, ROS/ROS2 or sensor synchronization work, and CI-for-robots style evaluation loops.