Design, train, and fine-tune models across NLP, computer vision, and multimodal pipelines, including adapting foundation and embedding models to DV's policy definitions
Own the full lifecycle: data selection -> labeling strategy -> training -> evaluation -> deployment
Develop strategies for efficient data curation and labeling (active learning, auto-labeling, LLM-based data generation, sampling under scale)
Improve model quality (precision/recall) while balancing cost, latency, and scale
Build rigorous evaluation methodology: test set design, data leakage prevention, and statistically sound comparisons
Apply modern AI approaches (LLMs, embeddings, foundation models, agentic workflows) to real production problems
Act as a technical leader for the team: mentor other Data Scientists, review approaches, and build reusable tools and playbooks that scale the team's output
Work closely with ML Engineering, Product, and Policy to translate ambiguous requirements into scalable systems
Who You Are
6+ years of hands-on experience in Machine Learning / Deep Learning
MSc or PhD in Computer Science, Statistics, Mathematics, or a related quantitative field
Strong background in NLP and/or Computer Vision
Track record of owning ML systems end-to-end, from problem framing and data strategy through deployment and monitoring, at scale
Strong understanding of real-world trade-offs (accuracy, cost, latency)
Able to own ambiguous problems independently, from framing through delivery
What We’re Looking For
Hands-on experience with deep learning frameworks (PyTorch / TensorFlow)
Hands-on experience fine-tuning and deploying deep learning models (e.g., transformers) to production
Experience working with large datasets and model evaluation pipelines
Advantages
Experience fine-tuning embedding models or foundation models for domain-specific tasks
Experience leveraging LLMs and other foundation models to extend team capabilities (e.g., labeling, data generation, evaluation)
Experience with multimodal systems (vision + text + audio)
Experience with multilingual NLP
Experience with active learning or data-centric AI
Experience mentoring other data scientists or leading technical projects
Why This Role
You will work on one of DV's most critical problems: building best-in-class classification systems under real-world constraints of scale, cost, and latency. This role offers deep technical ownership, high business impact, and a clear path to technical or people leadership.