Simulation Engineer – Cognitive Twin (human)

NEURA Robotics; Bielefeld; Vor Ort; EUR 70.000 - 110.000; Vollzeit

Zusammenfassung

NEURA Robotics seeks a robotics simulation engineer to stand up and operate its twin simulations on the platform’s physics engine. You will configure rigid-body dynamics, sensors, and environment models so scenes run reliably and repeatably.

You will build headless scenes, run standardized benchmarks, and evaluate control, planning, and learning approaches against fidelity and performance metrics.

Qualifikationen

• University degree or equivalent hands-on experience in robotics or related field.
• Several years in robotics simulation, benchmarking, or model evaluation.
• Hands-on with real-time 3D / game-engine based simulation environments.
• Understanding of multi-body dynamics, kinematics, and contact physics.
• Strong C++ (C++17) and CMake; Python for tooling.

Aufgaben

• Set up and operate twin simulation capabilities atop the platform's physics engine.
• Build and parameterize physics scenes and headless setups; wire in sensors/robots.
• Benchmark control, planning, and learning approaches using defined metrics.
• Conduct sim-to-real checks and calibrate simulation parameters.
• Develop ingest/export pipelines for 3D formats and robot descriptions.
• Own the simulation-backend scene definition mapping and real-time data streaming.
• Ensure reproducibility with clean configurations and benchmark reports.

Kenntnisse

C++17

CMake

Python

Experiment management

Real-time simulation

Multi-language codebase

Ausbildung

Bachelor's/Master's in Robotics/Mechanical/Electrical Engineering/CS/Mechatronics

Tools

MuJoCo

NVIDIA Isaac Sim

PyBullet

GPU-accelerated / distributed simulation

Scene-interchange tooling

Jobbeschreibung

Your mission & challenges

Simulation enablement: You stand up and operate the twin's simulation capabilities on top of the platform's physics engine — configuring rigid-body dynamics, contacts, and articulations, and bringing up robot, sensor, and environment models so scenes run reliably and repeatably.

Scene & content bring-up: You build and parameterize physics scenes and headless setups, wire in sensors and robots, and make simulation features accessible from the browser and the API.

Benchmarking & evaluation: You build standardized headless benchmark scenes and test cases, run them in the loop, and evaluate control, planning, and learning approaches against clearly defined fidelity and performance metrics — step rate, parallel instance count, latency, memory.

Sim-to-real checks: You help compare twin behavior against real-robot logs, surface obvious sim-to-real gaps (e.g. dynamics, sensor noise, contact behavior), and support calibration of simulation parameters as the fidelity bar takes shape.

Interoperability: You build and maintain ingest and export pipelines for common 3D and robot-description formats and own their round-trip fidelity, so scenes move cleanly in and out of the twin.

Platform integration: You own the simulation-backend side of the scene definition format — how a composed scene maps to the engine's runtime representation and how state is emitted over the binary real-time stream — together with Integration, Frontend, and AI/ML.

Reproducibility & quality: You keep scenarios, configurations, and results cleanly versioned and documented, set modeling conventions and acceptance criteria, and produce structured benchmark reports.

What we can look forward to

• A university degree (Bachelor's or Master's) in Robotics, Mechanical / Electrical Engineering, Computer Science, Mechatronics, or a related field — or equivalent hands-on experience.
• Several years of relevant experience in robotics simulation, benchmarking, or model-based evaluation, in research or industry.
• Hands-on expertise with a modern real-time 3D / game-engine-based simulation environment and a production physics engine (e.g. PhysX) — configuring rigid-body dynamics, articulations, and contacts — and confident robot / environment modeling with standard robot-description and scene-interchange formats.
• A working understanding of multi-body dynamics, kinematics, and contact physics — enough to configure, operate, and validate models with confidence.
• Strong modern C++ (C++17) and CMake, with Python for tooling; comfortable in large, multi-language codebases.
• Experience defining objective metrics and statistically evaluating simulation results, plus structured experiment management and automation of simulation tests (batch runs, sweeps, CI-like workflows).
• Experience with — or strong interest in — sim-to-real transfer, and a good understanding of real robot hardware and its limitations.
• Familiarity with AI coding tools (e.g. Claude Code, Cursor, Copilot) and agentic orchestration for development.
• A plus: MuJoCo, NVIDIA Isaac Sim, or PyBullet; GPU-accelerated / distributed simulation; actuator and sensor modeling (IMUs, LiDAR, cameras, force-torque); scene-interchange format tooling.
• A product-development mindset — you care about delivering real value to users and adapt readily as product requirements evolve.
• A structured working style, team spirit, and precise technical communication. Perfect command of English; German is a plus.