Santander is looking to develop intuitive, responsible, and user-centered experiences powered by Artificial Intelligence (AI) and Large Language Models (LLMs) to drive trust and value for customers and colleagues in the financial services industry.
Requirements
- Basic understanding of how LLMs, NLP, or AI decision systems function.
- Preferred, hands-on experience designing prompts for GPT-4, Claude, or Gemini and exposure to Snowflake Cortex.
- Strong communication and collaboration skills across design, product, and technical teams.
- Understanding of enterprise guardrails for hallucination detection, fallback, and escalation
- Interest in ethical AI, human-centered design, and user trust in automated systems.
- Ability to translate complex technical behaviors into clear, helpful user experiences.
- A growth mindset and a willingness to experiment and learn in a fast-evolving space.
Responsibilities
- Develop frontend and backend components of AI-driven apps using modern web frameworks and APIs.
- Integrate LLMs and ML models into business applications using APIs.
- Build workflows and interfaces that support natural language queries, report generation, summarization, or recommendation systems.
- Implement RAG pipelines using vector stores and enterprise content sources.
- Ensure performance, security, and scalability of cloud-native AI applications you build.
- Collaborate with prompt engineers to test and refine LLM outputs in real-time user contexts.
- Instrument applications for observability, error tracking, and user feedback to drive continuous improvement.
Other
- Master’s degree in a quantitative area (Computer Science preferred) or a terminal degree in any field.
- 3 years of relevant professional (or research) experience
- Frequent minimal physical effort such as sitting, standing and walking is required for this role.
- Depending on location, occasional moving and lifting light equipment and/or furniture may be required.
- Ability to work in a team environment and collaborate with cross-functional teams.
