Senior Machine Learning Engineer – Multimodal Data

This is a fully remote position, open to applicants in Austria.

📋 Description

• Design and develop data pipelines for agent training, including tasks such as collection, filtering, deduplication, formatting, and versioning across various sources like text, images, and multimodal data.

• Construct and oversee infrastructure to ensure scalable data loading, storage, and retrieval (S3, distributed systems, streaming pipelines).

• Work collaboratively with research scientists to convert research needs into specific data specifications, adapting as experiments uncover new requirements.

• Generate evaluation datasets and benchmarks in partnership with researchers, curating task distributions that highlight actual failure modes.

• Create tools for dataset construction, which includes human annotation workflows, synthetic data generation, and preference data collection for RLHF/DPO-style training.

• Take ownership of data quality by developing validation frameworks, monitoring for drift and contamination, and setting standards that ensure datasets are reliable and reproducible.

• Thoroughly document datasets, including their provenance, known limitations, intended use cases, and versioning history.

• Implement extensive test coverage for data pipelines and ML workflows to guarantee reliability and identify regressions early.

• Enhance codebase quality through code reviews, refactoring, and establishing engineering best practices that promote sustainable research velocity.

• Contribute to team roadmaps by identifying data bottlenecks and suggesting solutions that facilitate research progress.

⛳️ Requirements

• Proficient software engineering skills in Python, with experience in building production-grade data pipelines and ML DevOps.

• Practical expertise in prompt engineering, including designing, testing, and refining prompts for consistent LLM/VLM outputs.

• Experience with ML data workflows, including large-scale data processing and loading (Ray or similar), data versioning, and training format considerations (tokenization, batching, sharding).

• Hands-on experience with data pipelines for large-scale distributed ML training runs.

• Familiarity with annotation tools and human-in-the-loop data collection (such as Label Studio or internal systems).

• Strong understanding of ML training requirements, recognizing what constitutes "good data" for LLM/VLM fine-tuning and anticipating potential downstream issues.

• Experience in loading and writing large datasets to and from cloud infrastructure (AWS) and distributed storage systems.

• Excellent communication skills, enabling effective collaboration with researchers to define ambiguous problems and translate needs into actionable plans.

• A collaborative mindset, comfortable with taking ownership and iterating quickly.

🏝️ Benefits

• Equity packages - we want our success to be yours too.

• An inclusive parental leave policy that supports all parents and caregivers.

• An annual Vibe & Thrive allowance to enhance your wellbeing, social connections, office setup, and more.

• Flexible leave options that empower you to contribute positively, take time to recharge, and support your personal needs.