Full-Stack AI Engineer / Data Scientist – Agentic Systems

About the role

Core technology scope

• Work with multimodal foundation models, agentic workflows, Agent-to-Agent communication, Model Context Protocol, Retrieval-Augmented Generation, and MLOps subsystems.
• Use AWS and multicloud services with vector, graph, traditional databases, and data lakes to build scalable AI solutions.
• Use advanced Python, occasional TypeScript, LangChain or similar frameworks, state machine logic, Docker, CI/CD, infrastructure as code, and automated testing.
• Apply statistics, data manipulation, synthetic data generation, feature engineering, prompt engineering, and model optimization.

Agentic and full-stack development

• Design and build agentic systems, state machines, and search-based conversational systems for complex business problems.
• Develop multimodal foundation model workflows for text, audio, and video.
• Implement MCP servers and clients to standardize context exchange between agents, data sources, and external tools.
• Write production-ready Python and occasional TypeScript code with testing and documentation.
• Manage containerization, CI/CD, and reliable cloud-native solutions across AWS and multicloud environments.

Data science and algorithms

• Perform exploratory data analysis to identify patterns, bias, and data quality issues.
• Use statistical and visualization techniques for feature engineering, model selection, and application optimization.
• Curate and preprocess diverse datasets to improve LLM effectiveness and reduce bias.
• Design, customize, optimize, and fine-tune LLM-based and traditional AI algorithms.
• Optimize inference speed, latency, caching strategies, and resource usage.

Evaluation and observability

• Conduct experimentation and A/B testing using automated metrics, retrieval accuracy measures, and human rating frameworks.
• Implement monitoring and observability for pipeline behavior, drift detection, anomalies, and real-time algorithms.
• Translate technical results into clear insights for stakeholders and support data-driven decisions.

Experience and qualifications

• 7+ years of AI/ML engineering and data science experience, including generative AI, agents, and classical machine learning.
• 3+ years of hands-on experience with generative large language models and agents.
• Proven end-to-end MLOps experience and model deployment at scale.
• Proven experience using AI coding agents in an agentic software development lifecycle.
• A relevant bachelor's, master's, engineering, or doctoral degree, or equivalent.
• English communication at C1+ level.

Capabilities

• Strong analytical and problem-solving skills for complex engineering and statistical challenges.
• A strong sense of ownership and ability to define architectural patterns and step into a Tech Lead role when needed.
• Ability to consult with enterprise stakeholders on technology approaches.
• Understanding of bias, fairness, hallucination mitigation, and responsible AI deployment.
• Team-oriented, proactive, collaborative, and able to communicate complex findings clearly.

Why join Roche

• Work on cutting-edge agentic AI technologies in a highly visible Roche Digital Technology function.
• Take ownership of complex, high-impact end-to-end AI systems.
• Join a collaborative culture focused on technical excellence and rapid innovation.