You will dedicate around 50% of your time to professionalising an internal platform created to run a multi-asset, multi-theme sustainability impact portfolio for the Wealth Management division. The platform is a Python and LLM-based application combining a PostgreSQL database, LLM-based agents and human-in-the-loop workflows for investment assessment, portfolio analytics and reporting. Your mission will be to help evolve the platform towards enterprise standards – cloud infrastructure, multiple users, security and governance – while continuing to extend what it can do.
The rest of your time will be dedicated to sustainability data and automation initiatives across the Group Sustainability Office. This can include developing sustainability integration tools for our investment teams, user-friendly dashboards for the GSO and ad-hoc data analysis to support ongoing GSO projects.
This is a hands-on role for someone equally at ease in a database, in a prompt, and in a meeting with the bank’s engineering, data, infrastructure and security teams.
You will be in charge of:
- Build and evolve the platform: Co-design data pipelines, LLM-based agents, analytics and reporting features with the business line, and build user-friendly GSO dashboards in coordination with Finance and Data teams.
- Own and understand the data: Identify, source and define required data, maintaining a single source of truth across schema, ingestion, data quality, versioning, auditability and large complex datasets.
- Industrialise with strong engineering: Move prototypes to the bank’s cloud infrastructure with secure, monitored multi-user access, and enforce solid testing, documentation, deployment and audit practices.
- Ensure trustworthy, cutting-edge AI: Implement evaluation, guardrails and human review for explainable outputs, while assessing and responsibly deploying new LLM and agent capabilities.
- Act as the interface: Translate between business and technical teams and navigate engineering, infrastructure, security and compliance stakeholders to get decisions made.