DS Consultant – Data Mgmt & Applied AI (NCS/Job/ 1871)

• Lead or support data annotation projects, focusing on imbalanced datasets and classification quality.
• Design data labeling workflows that support prompt fine-tuning and intent detection improvements.
• Support data augmentation and synthesis to expand training sets for underrepresented categories.
• Perform embedding generation, vectorization, and feature extraction for downstream model use.
• Apply dimensionality reduction and clustering to support cohort modeling and personalization efforts.
• Contribute to tools and processes that generate reusable feature sets for prompt APIs and Copilot skills.
• Fine-tune small/medium scale custom models using annotated datasets and embedding-based features.
• Build unit/offline test frameworks for model behavior validation, especially around intent detection and prompt understanding.
• Conduct prompt-level accuracy evaluations and maintain version-controlled test scripts.
• Develop lightweight custom models tailored to specific use cases such as Live Notes, Meeting Summarization, and Skill Invocation.
• Work with engineering teams to flight models across controlled releases, enabling A/B comparison and performance measurement.
• Document findings, test protocols, and improvement paths across releases.

Required Experience

• 3+ years of experience in applied data science, ML/NLP workflows, or model operations.
• Proven expertise in Python, Pandas, scikit-learn, PyTorch/TensorFlow, and embedding frameworks.
• Hands-on experience with data labeling platforms, annotation QA, or prompt training pipelines.
• Deep understanding of feature engineering for NLP or structured intent-based models.
• Prior work in unit testing models, offline evaluation, or custom model deployments preferred.

Soft Expectations

• Strong communication and documentation skills.
• Ability to work independently with clear weekly deliverables.
• Comfort in engaging with cross-functional teams including engineers, product managers, and peer data scientists.
• Time zone alignment with Europe is preferred for coordination with the Prague team.