Adopt the latest technologies and trends in NLP to your platform
Develop LLM-based agents capable of performing function calls and utilizing tools such as browsers for enhanced data interaction and retrieval.
Experience with Reinforcement Learning from Human Feedback (RLHF) methods such as Direct Preference Optimization (DPO) and Proximal Policy Optimization (PPO) for training LLMs based on human preferences.
Design, develop, and implement an end-to-end pipeline for extracting predefined categories of information from large-scale, unstructured data across multi-domain and multilingual settings
Create a robust semantic search functionality that effectively answers user queries related to various aspects of the data
Use and develop named entity recognition, entity-linking, slot-filling, few-shot learning, active learning, question/answering, dense passage retrieval and other statistical techniques and models for information extraction and machine reading
Deeply understand and analyze our data model per data source and geo-region and interpret model decisions
Collaborate with data quality teams to define annotation tasks, metrics, and perform qualitative and quantitative evaluation
Utilize cloud infrastructure for model development, ensuring seamless collaboration with our team of software developers and DevOps engineers for efficient deployment to production
4+ years of experience as a data scientist (or 2+ years with a Ph.D. degree)
Master's or Ph.D. in Computer Science, Artificial Intelligence, Computational Linguistics, or a related field.
Strong theoretical knowledge of Natural Language Processing, Machine Learning, and Deep Learning techniques.
Proven experience working with large language models and transformer architectures, such as GPT, BERT, or similar.
Familiarity with large-scale data processing and analysis, preferably within the medical domain.
Proficiency in Python and relevant NLP libraries (e.g., NLTK, SpaCy, Hugging Face Transformers).
Experience in at least one framework for BigData (e.g. Ray, Spark) and one framework for Deep Learning (e.g. PyTorch, JAX)
You can work remotely anywhere in the UK, The Netherlands or Spain and you have to be a resident of one of the aforementioned countries and be legally authorized to work there without requiring Veeva’s support for visa or relocation. If you do not meet this condition, but you think you are an exceptional candidate please clarify it in a separate note and we will consider it.