Principal AI Scientist

RWS Group is seeking a Principal AI Scientist to join its rapidly expanding TrainAI division — our global center of excellence, for training, evaluating, and scaling the next generation of artificial intelligence systems.

TrainAI partners with some of the world's most influential technology companies to develop, refine, and optimize advanced AI models, ensuring they perform reliably across languages, domains, and real-world environments. In this role, you will define the scientific and technical strategy that underpins our AI training programs while personally building the prototypes, pilots, and production-grade tools that bring that strategy to life. You will directly contribute to the evolution of state-of-the-art large language models and intelligent systems — not just through research direction, but through hands-on engineering and experimentation.

As a Principal AI Scientist, you will combine deep machine learning expertise with research leadership and builder-mindset execution. You will drive the design of robust ML methodologies, architect large-scale evaluation and benchmarking frameworks, build working MVPs that demonstrate new service capabilities, and collaborate closely with cross-functional teams to translate research advancements into scalable, production-ready AI capabilities used globally.

The TrainAI division is shaping the future of artificial intelligence by uniting human insight with machine learning at enterprise scale. Our teams enable AI to better understand language, context, and culture — supporting some of the world's most sophisticated technologies.

In this role, you will play a pivotal part in defining how AI systems learn, reason, and operate in the real world — and you will build the tools and systems that prove it works.

About RWS TrainAI Data Services

Training or fine-tuning artificial intelligence (AI) requires data. LOTS of data. However, not just any data will do – our clients need responsible AI data that’s targeted, accurate and reliable to ensure machine learning (ML) success. But preparing AI training data is a monumental task that can take up the vast majority of AI project time, leaving AI teams with precious little time to focus on developing, deploying and evaluating ML models. RWS Train AI helps our clients address this challenge head-on.

With a seamless blend of technological understanding and human intelligence, RWS Train AI provides complete, end-to-end data collection and content generation, data annotation or labelling, human-in-the-loop data validation and generative AI data services for all types of AI, in any language, at any scale, based on the principles of responsible AI. Today, Train AI supports four of the world’s top five technology companies, enhancing the performance of their generative AI applications by providing services such as prompt engineering, response refinement and red teaming with locale-specific domain experts across a broad range of topic areas and educational levels.

For further information, please visit: RWSTrainAI

Key responsibilities

Advance AI Data Services

• Lead the development of innovative approaches to training, evaluating, and improving large language models and multimodal AI systems.
• Design methodologies for large-scale data annotation, human-in-the-loop learning, and model alignment.
• Apply advanced techniques in NLP, reinforcement learning, and generative AI to improve model accuracy, safety, and performance.
• Conduct hands-on experimentation — running model evaluations, fine-tuning approaches, and benchmarking studies that generate publishable insights and inform client engagements.

Build Pilots, MVPs & Automated Solutions

• Personally architect and build working prototypes that demonstrate new TrainAI service capabilities — such as automated evaluation pipelines, hybrid human-AI quality systems, LLM-as-judge frameworks, and domain-specific benchmarking tools.
• Develop internal automation tools that improve annotation quality, reduce delivery costs, and increase operational efficiency across TrainAI programs.
• Build proof-of-concept solutions for client-facing opportunities — translating technical feasibility into tangible demos that support sales conversations and RFP responses.
• Own the end-to-end technical development lifecycle for pilot projects: scoping, rapid prototyping, testing, iteration, and handoff to engineering for productionization.
• Create reusable frameworks, libraries, and toolkits that codify TrainAI's technical IP and enable repeatable delivery at scale.

Shape the TrainAI Research Agenda

• Define scientific priorities and research initiatives that push the boundaries of AI training and evaluation.
• Explore emerging techniques in AI alignment, prompt engineering, agentic systems, and model optimization.
• Translate cutting-edge research into scalable solutions deployed across TrainAI programs.
• Publish research papers, author white papers, and represent TrainAI at industry confere