Principal Data Scientist, Generative Recommendations

Select how often (in days) to receive an alert: Create AlertPrincipal Data Scientist, Generative Recommendations & Agentic Orchestration- Spain RemoteDate: Sep 27, 2026Location: Barcelona, B, ES, 08019Company: VistaWho We AreAt Vista, we are building an industry-leading Customer Relevance Platform to scale personalization across products, design templates, and marketing messages.

Es esencial asegurarse de que cumple con los requisitos como solicitante para este puesto; por favor, lea atentamente la información a continuación.

We are moving away from 'model sprawl' toward a unified architecture centered on fmX, a foundation model that learns shared embeddings for all relevant entities for Vista (users, products, marketing messages, templates, etc.).Our strategic vision focuses on delivering a "Context-aware Customer Experience." To achieve this, we are prioritizing the evolution toward Intent-Based Personalization, capturing real-time journey signals, and infusing our recommendations into Agentic Experiences.

This means equipping AI assistants with an Agentic Reasoning Layer that sits on top of our fmX rankers, enabling personalized recommendations to fit seamlessly into agentic workflows and user experiences.What You Will DoAs the Principal Data Scientist, you will drive the generative, intent-based, and agentic evolution of our platform.

A critical part of your role will be to act as the primary bridge between traditional data science practices (ranking, embeddings, deep learning) and advanced LLM / Agentic reasoning layers.

You will partner directly with another Principal Data Scientist to co-lead our research agenda, accelerating our strategic progress and elevating the capabilities of the wider Data Science team.Bridging Data Science & Agentic Experiences (Key Responsibility): Act as the primary technical bridge between core ML/DS modeling and modern LLM application layers.

Translate classical recommendations, graph embeddings, and fmX outputs into robust reasoning abstractions for use in agentic workflows.Agentic Reasoning & Recommendations: Lead the evolution of recommendations into the Generative AI era.

Design and build the agentic reasoning layer that wraps around our fmX models and rankers.

Implement patterns that allow models to self-correct, reason, and contextually explain recommendations to customers.Tool-Use & MCP Architecture: Architect and build the ML pathways and Model Context Protocol (MCP) servers to expose core recommendation models and data as modular tools for AI agents.

This allows conversational shopping experiences to dynamically query, rank, and synthesize candidates in real time.Intent-Based Personalization: Develop an algorithmic intent layer that translates real-time browsing and session signals into structured, context-aware customer intents to feed directly into the fmX-based recommendation pipeline.Broad Personalization & Segmentation: Build advanced propensity models and leverage behavioral traits to drive actionable, business-aligned customer segmentations and dynamic personas.AI Leadership & Strategy: Act as Vista's AI ambassador, influencing product, engineering, and business leadership to adopt a model-driven, agent-first approach.

Ensure models integrate cleanly with delivery systems without hurting Core Web Vitals or caching efficiency.Your QualificationsAdvanced degree (Ph.D.

or Master's) in Computer Science, Applied Mathematics, Statistics, or a closely related quantitative field.8+ years of experience building and deploying machine learning models and deep learning systems into high-scale customer experiences.Hands-on experience and proactive experimentation with Agentic AI, AI Agent Orchestration frameworks (e.g., LangGraph, Strands, etc.), and building reasoning layers (such as advanced chain-of-thought, self-correction, or validation loops).Demonstrated experience or active thinking/experimentation in building Model Context Protocol (MCP) servers to bridge data/ML systems with LLM orchestrators in a production context.Deep expertise in Deep Learning, Large Language Models (LLMs), and Reinforcement Learning (especially applied to real-time feedback loops or personalized ranking).Proven track record of building propensity models and customer segmentation models on large-scale behavioral data.Expert-level proficiency in Python, PyTorch/TensorFlow, and SQL.Strong architectural understanding of how ML models interact with both backend pipelines (Databricks, Snowflake, streaming features) and frontend delivery systems (UI rendering, caching strategies, latency management).Very strong communication and stakeholder management skills, with a proven ability to translate complex statistical and algorithmic concepts into clear business value to influence non-technical leaders and align cross-functional engineering teams.Why You'll Love Working HereThere is a lot to love about working at VistaPrint