Senior Machine Learning Scientist

Senior Machine Learning Scientist

The Senior Machine Learning Scientist is responsible for building and evaluating GenAI‑ and LLM‑powered solutions and AI agents that improve post‑booking customer experience, including recommendations, customer service, and trip management. The role owns end‑to‑end ML and GenAI projects—from problem framing and data preparation through model/agent design, orchestration, deployment, and continuous evaluation. The scientist applies deep expertise in applied ML, Generative AI, and rigorous experimentation to design robust evaluation frameworks (A/B tests, offline metrics, qualitative assessments) that ensure agents are safe, effective, and aligned with business goals.

The position partners closely with product, engineering, and operations while mentoring junior scientists and helping define best practices for AI agent development and evaluation.

• Own the end‑to‑end ML lifecycle for medium‑to‑large projects: from problem framing and ideation through research, prototyping, deployment, and post‑launch monitoring.
• Design robust, scalable ML systems (batch and/or streaming) in partnership with engineering, including data pipelines, feature computation, and model serving.
• Translate ambiguous business problems into well‑defined ML problems with clear success metrics and validation strategies.
• Develop, evaluate, and iterate on supervised, unsupervised, and deep learning models for prediction, recommendation, and optimization.
• Apply causal inference and experimental design (A/B testing) to accurately measure impact and guide decision‑making.
• Read and apply relevant academic and industry research to improve model architectures, training strategies, and evaluation methods.
• Contribute to defining best practices for experimentation and modeling within the team; help raise the technical bar for ML development.
• Build and iterate on models and applications leveraging GenAI / LLM technologies (e.g., OpenAI, Hugging Face, Anthropic, Gemini) for customer support, content generation, and workflow automation.
• Use prompting, retrieval‑augmented generation, and tool/function‑calling patterns to integrate LLMs into production systems.
• Explore and prototype advanced ML techniques (e.g., reinforcement learning, sequence modeling, transformers) where they can provide clear business value.
• Design end‑to‑end modeling approaches, including data selection, feature engineering, algorithm choice, training procedures, and evaluation.
• Apply statistical rigor in analyzing experiments and observational data; quantify uncertainty, trade‑offs, and model risk.
• Define and monitor offline and online metrics that faithfully reflect business goals (e.g., customer satisfaction, cost‑to‑serve, operational efficiency).
• Partner closely with product managers, engineers, analysts, and operations to understand requirements, define roadmaps, and align on priorities.
• Communicate complex technical concepts in a clear, concise way to technical and non‑technical stakeholders.
• Build intuitive dashboards and visualizations to explain model behavior, experiment results, and business impact.
• Lead cross‑functional projects involving multiple partners (e.g., product, engineering, operations), driving them from conception to measurable impact.
• Manage project scope, timelines, and communication, proactively surfacing risks and trade‑offs.
• Mentor junior scientists and engineers on modeling approaches, experimentation, and analytical problem solving.

• PhD in a quantitative field (e.g., Computer Science, Statistics, Mathematics, Physics, Economics, Operations Research) and ~3+ years of industry experience; or Master’s degree in a quantitative field with ~5+ years of relevant industry experience.
• Proven track record of building and deploying ML models that meaningfully impact business metrics in a production environment.
• Strong knowledge of machine learning theory and practice (supervised learning, representation learning, ranking/recommendation, deep learning).
• Solid grounding in statistics, experimental design (A/B testing), and basic causal inference; comfortable designing and analyzing online…