Data Scientist, Multiverse Computing

• We are seeking a skilled and experienced AI Evaluation Data Scientist with a strong technical background in Generative AI to join our team
• Master's or Ph.D. in Computer Science, Machine Learning, Data Science, Physics, Engineering, or related technical fields, with relevant industry experience
• Solid hands-on experience (3+ years for mid-level, 5+ years for senior) working as a Data Scientist, ML Engineer, or Research Scientist in applied AI/ML projects deployed in production environments
• Strong background in evaluation of machine learning systems, ideally with experience in LLMs, RAG pipelines, or multi-agent systems
• Proven ability to design and implement evaluation methodologies that go beyond static benchmarks, capturing real-world task success, reasoning, and robustness
• Hands-on experience with dataset creation and curation (including synthetic data generation) for training and evaluation
• Proven experience with agent-based architectures (task decomposition, tool use, reasoning workflows), RAG architectures (retrievers, vector databases, rerankers), and orchestration frameworks (LangGraph, LlamaIndex)
• Strong problem-solving skills, with the ability to navigate ambiguity and design practical solutions to open-ended user or business needs
• Strong software engineering skills, with proficiency in Python, Docker, Git, and experience building robust, modular, and scalable ML codebases
• Familiarity with common ML and data libraries and frameworks (e.g., PyTorch, HuggingFace, LangGraph, LlamaIndex, Pandas, etc.)
• Experience with cloud platforms (ideally AWS)
• Excellent communication skills, with the ability to work collaboratively in a team environment, document and explain design decisions, experimental results, and communicate complex ideas effectively
• Fluent in English

Desirable

• Fixed-term contract ending in June 2026
• In this role you will have the opportunity to lead the design and implementation of evaluation frameworks to assess the performance of Generative AI systems before deployment in production, as well as working closely with cross-functional teams to turn outcomes into actionable insights to be integrated into our products
• You will have the opportunity to work on challenging projects and shape the future of Generative AI systems
• Design and lead the evaluation strategy for our Agentic AI and RAG systems, turning customer workflows and business needs into measurable metrics and clear success criteria
• Contribute to the end-to-end design of Agentic AI and RAG systems, injecting a data-and-evaluation perspective into retrieval strategies, orchestration policies, tool usage, and memory to solve complex, real-world problems across industries
• Develop task-based, multi-step evaluations that reflect how the different components of our systems (retrieval, planning, tool use, memory) perform in real-world scenarios across cloud and edge deployments
• Develop and refine rigorous evaluation frameworks that reflect real-world performance, going beyond model benchmarks to assess task success, reasoning capabilities, factual consistency, reliability, and user success metrics across diverse problem domains
• Build and maintain a reproducible evaluation pipeline, including datasets, scenarios, configs, test suites, versioned assets, and automated runs to track regressions and improvements over time
• Curate and generate high-quality datasets for evaluation, including synthetic and adversarial data, to strengthen coverage and robustness
• Implement and calibrate LLM-as-a-judge evaluations, aligning automated scoring with human feedback and ensuring fairness, robustness, and representativeness
• Perform deep error analyses and ablations to uncover failure patterns, maintain a taxonomy of failure modes (reasoning, grounding, hallucinations, tool failures), and provide actionable insights to engineers to improve model and system performance
• Partner with ML specialists to create a data flywheel, where evaluation continuously informs new dataset creation, improvements on prompts, tool usage, model training, and system refinements, quantifying improvements over time
• Define and monitor operational metrics (latency, cost, reliability) to ensure evaluations align with production and customer expectations
• Maintain high engineering standards, including clear documentation, reproducible experiments, robust version control, and well-structured ML pipelines
• Contribute to team learning and mentorship, guiding junior engineers and sharing expertise in LLM development, evaluation, and deployment best practices
• Participate in code reviews, offering thoughtful, constructive feedback to maintain code quality, readability, and consistency

• We are looking to fill this role immediately and are reviewing applications daily. Expect a fast, transparent process with quick feedback