Applied Data Scientist

About the role

Spacebackend is seeking an Applied Data Scientist to develop AI‑driven models that translate technical interface specifications into digital‑twin representations, detect ambiguities, and enable autonomous spacecraft behavior. You will work at the intersection of machine learning, aerospace engineering, and software integration.

Key responsibilities

• Design and train custom transformer‑based models to ingest manufacturer interface specifications and generate structured digital‑twin schemas.
• Develop embedding, retrieval‑augmented generation and knowledge‑graph pipelines to ensure high‑fidelity mapping of hardware capabilities.
• Fine‑tune large‑scale models to detect missing or conflicting information in specifications and provide curated feedback to engineers.
• Prototype neural and hybrid architectures (e.g., CNNs, reinforcement‑learning policies, LLMs) that use digital‑twin representations to command and simulate spacecraft behavior.
• Lead hypothesis‑driven research, design experiments, collect and analyse data, and disseminate findings internally.
• Establish reproducible ML infrastructure including experiment tracking, model versioning and containerised training pipelines (MLflow, Docker, Kubernetes).
• Mentor junior researchers and interns, fostering knowledge sharing within an iterative R&D culture.

Required profile

• Eligible to work in Luxembourg without visa sponsorship.

Required skills

• Transformer‑based model design and training.
• Embedding methods, retrieval‑augmented generation, knowledge‑graph techniques.
• Large‑scale language model fine‑tuning for ambiguity detection.
• Neural architectures such as CNNs, reinforcement‑learning policies, and LLMs.
• MLflow, Docker, Kubernetes for reproducible ML pipelines.