2 days ago
Dublin, IrelandStaff+
Responsibilities
- Design and build language-based skills for enterprise AI assistants such as Microsoft Copilot Cowork and Claude Cowork.
- Translate programme management, governance, portfolio reporting, and development processes into precise, testable skill specifications.
- Design workflows across skills, agents, tools, and human decision points, including agent-to-agent integration patterns.
- Use Claude Code, GitHub Copilot, and AWS Kiro to author, test, version, and maintain skills, connectors, data preparation, and evaluation harnesses.
- Organize shared, trusted, versioned knowledge layers using approved connectors and MCP servers for sources such as SharePoint, Teams, and business systems.
- Build repeatable evaluation systems using real documents, expert reference answers, structured review, and AI-assisted grading checked by experts.
- Define safeguards and approval paths for skills that draft, update, or send information, including least privilege, reversibility, previews, and audit trails.
- Ensure acting systems meet GxP, data privacy, and Responsible AI requirements in collaboration with Quality, Compliance, and DDIT.
- Embed with SPPO and Process Innovation teams to identify use cases, deploy skills, and measure adoption and impact.
- Coach early-career scientists and contribute to evaluation standards, publications, conferences, and communities of practice.
- Run a reusable skill-building factory with intake, triage, shared components, and a maintained register of systems, owners, evaluation status, and value.
Requirements
- At least 6 years of experience as a software, data, or ML engineer or data scientist building reliable systems.
- Recent demonstrable experience building skills, agents, or workflows for AI assistants such as Microsoft Copilot or Claude, including evaluating their performance.
- Experience designing structured evaluations, test cases, expert review processes, evidence-based quality judgments, and precise specifications.
- Ability to read and write code in Python or a similar language and design data and workflow flows through systems.
- Experience with data models, APIs, connectors, knowledge layers, MCP servers, and agent-to-agent integration patterns.
- Bachelor-level degree in computer science, engineering, or a quantitative or life science field with strong technical depth.
- Experience in drug-development governance, programme management, portfolio processes, regulated GxP environments, process improvement, Lean, Six Sigma, or design thinking is desirable.
- Experience using Claude Code, GitHub Copilot, or AWS Kiro is desirable; postgraduate education is an advantage.
Benefits
- Hybrid working in Dublin, Ireland; relocation is not offered.
- Insurance plans, retirement plans, wellbeing resources, and global recognition programmes.
- Flexible and hybrid working options where possible.
- Minimum of 14 weeks paid parental leave.
- Potential eligibility for a performance-based bonus in addition to base salary.
About Novartis
Novartis is an innovative medicines company. Every day, working to reimagine medicine to improve and extend people’s lives so that patients, healthcare professionals and societies are empowered in the face of serious disease. Our medicines reach more than 250 million people worldwide. Find out more at https://www.novartis.com See our community guidelines: https://go.novartis.social/3Nboxki
