About the Role
Airbnb is dedicated to creating a world where anyone can belong anywhere. Travel should feel safe, and AirCover delivers on that promise through Guest Travel Insurance (GTI), offering guests peace of mind at the moment of booking and throughout their trip. As a Data Scientist on AirCover, you will work at the intersection of insurance, personalization, and machine learning, building intelligent systems that help the right guest discover the right coverage at the right moment. You will join a tight-knit, high-output DS team that manages one of Airbnb’s most experiment-dense personalization roadmaps, partnering daily with product, engineering, operations, and legal to ship work that directly affects guest trust and revenue.
The Difference You Will Make
We are looking for a machine learning expert excited to own hard problems end-to-end, from prototype to production. You will have direct scope to contribute and lead across:
- Package personalization & ML-based recommendation: Evolve rule-based guest segmentation into a full ML recommendation system that surfaces the right insurance (e.g., trip cancellation, accidental damage coverage, on-trip protection) to each guest based on purchase intent, trip attributes, listing signals, and user history.
- Content personalization: Build models that rank and select benefit messaging for each guest, deciding which coverages to highlight, in what order, and with what framing, drawing on learnings from segmentation experiments and LLM-assisted content prototyping.
- Intent modeling: Develop and productionize ML models (from gradient-boosted trees to deep learning) that predict a guest’s likelihood to value specific coverages, using structured booking data and unstructured signals.
- Journey understanding and optimization: Leverage reinforcement learning to personalize across the user journey, with understanding of user preferences on entry point, price, notification frequency, and trip characteristics.
- High-velocity experimentation: Design and run adaptive experiments to maximize learning within tight traffic constraints; sequence ERFs strategically to keep the personalization roadmap moving.
A Typical Day
- Dig into experiment results to surface high-impact personalization opportunities; translate findings into crisp scientific problem formulations that balance rigor with speed-to-learning.
- Work closely with product managers, engineers, operations, legal, and privacy partners to align on ML requirements, de-risk design decisions, and gather requirements on explainability and compliance.
- Hands-on develop, evaluate, and ship ML models and data pipelines at scale (batch and real-time, structured and unstructured) using Airbnb’s paved-path tooling and AI native mindset.
- Prototype and iterate quickly: turn a new idea into a working model in a prototype, get early signals from an experiment, then productionize what works.
- Present findings and proposals at team reviews and to technical, product, and executive stakeholders, making complex ML results legible without oversimplifying, and generating conviction on the roadmap ahead.
- Stay current with the research community; draw on state-of-the-art advances in recommendation systems, LLMs, and personalization to raise the bar for the team's output. Occasionally publish externally or present at conferences to advance Airbnb’s scientific standing.
Your Expertise
- 5+ years of relevant industry experience (e.g., ML scientist, tech lead, junior faculty) and a Master’s degree or PhD with 2+ years in a relevant field.
- Proven hands-on experience building and shipping personalization and recommendation systems at scale: strong intuition for feature engineering, user modeling, and the full ML lifecycle (training, serving, monitoring, iteration). Experience with LLMs, Computer Vision or content-understanding topics is a strong plus.
- Strong fluency in Python and SQL; hands-on experience with TensorFlow or PyTorch, Airflow, and a data warehouse environment.
- Deep understanding of ML algorithms (gradient-boosted trees, deep learning, optimization) and experiment design, including A/B testing, multi-armed bandits, and the practical constraints of running experiments at scale. Causal inference skills are a plus.
- Exceptional communicator: you can make complex ML work legible to engineers, product managers, legal, and executives alike—written and verbal. You treat communication as a core part of the job, not an afterthought.
- Self-directed and passionate: you are energized by a fast-moving environment where there are always more good ideas than time; you hold yourself to a high standard without being asked, take initiative to unblock yourself, and find genuine satisfaction in shipping things that matter to guests.
- Product-oriented mindset: you keep the guest experience at the center of technical decisions and bring conceptual and innovative thinking to how you frame and solve problems. Publications or presentations in recognized venues are a plus.