Airbnb's loop is built around a two-sided marketplace and a company-wide experimentation practice, so every question carries a host perspective and a guest perspective at once. System design leans on experimentation platforms, search-ranking pipelines, and pricing analytics, and a dedicated core-values interview with veto power evaluates alignment with the mission alongside the technical rounds.
First contact to offer, stage by stage, with how long each round runs and roughly when it lands.
The loop stage by stage, from recruiter call to offer.
Initial conversation about your experience and interest in Airbnb. The recruiter evaluates your background with data platforms, pipeline orchestration, and experimentation infrastructure. Airbnb has one of the strongest data cultures in tech, so they look for candidates who understand how data drives product decisions and can articulate the role of data engineering in enabling that culture.
SQL problems set in a marketplace context: bookings, listings, reviews, search interactions. Airbnb phone screens focus on aggregation, window functions, and multi-step analytical queries. The interviewer evaluates clarity of thought and how you handle ambiguity in problem definitions. You may need to ask clarifying questions about business logic before writing queries.
5 rounds covering SQL deep dive, system design, coding (Python), data modeling, and a core values interview. System design at Airbnb emphasizes experimentation platforms, search ranking data pipelines, and pricing analytics. The core values interview evaluates alignment with Airbnb's mission and how you champion the data culture. Data modeling often involves marketplace schemas that support both operational and analytical workloads.
Our prediction of the question mix by domain and difficulty for this company's data engineer loop, from live listings and interview reports.
The domain and difficulty mix we predict for a Airbnb data engineer loop, across 13 problems. It updates as more Airbnb data lands.
Reported by candidates from real loops, tagged by domain, round, level, and year. Expand for what the round is scoring.
Python coding question from Airbnb Data Engineer virtual onsite. Expected approach: for each triplet of sides (a, b, c) from the list, check the triangle inequality: a + b > c AND a + c > b AND b + c > a. Brute force: use itertools.combinations(sides, 3) to generate all triplets and count valid ones — O(n^3). Optimized: sort the array first, then for each pair (i, j), use binary search to find the count of valid third sides — O(n^2 log n). Edge cases: duplicate side lengths should still form distinct triangles if at different indices, and degenerate triangles (a + b == c) are typically…
Write a function build_file_tree(paths) that takes a list of file path strings and returns a nested dictionary representing the directory structure. Files (leaf nodes) should have a value of None. Paths are always relative (no leading '/'), contain no '..' or '.' segments, and have no trailing slashes. Example: paths = ['src/main.py', 'src/utils/helper.py', 'README.md', 'src/utils/config.py'] Output: { 'src': { 'main.py': None, 'utils': { 'helper.py': None, 'config.py': None } }, 'README.md': None } paths = ['a/b/c/d.txt'] Output: {'a': {'b': {'c': {'d.txt': None}}}} paths = [] Output: {}…
Write a function build_tree(pairs) where each pair is a tuple (parent, child). Return the tree as a nested dictionary. Assume a single root node (one value that appears as a parent but never as a child), no cycles, and each child has exactly one parent. Leaf nodes map to an empty dict. Example: pairs = [('A', 'B'), ('A', 'C'), ('B', 'D'), ('B', 'E'), ('C', 'F')] Output: {'A': {'B': {'D': {}, 'E': {}}, 'C': {'F': {}}}} pairs = [('root', 'a'), ('root', 'b')] Output: {'root': {'a': {}, 'b': {}}} pairs = [('x', 'y'), ('y', 'z')] Output: {'x': {'y': {'z': {}}}} Edge cases: single pair, linear…
Airbnb onsite data modeling round. Candidate presented with 5 business requirements and asked to develop a complete data model. Expected entities: transaction table (fact), users, hosts, properties, price table, reviews, geography, and time bucket tables. Interviewer probed on primary/foreign key decisions, normalization vs denormalization trade-offs, how to partition data for scale, and how to handle slowly changing dimensions for pricing. Follow-up: write SQL queries against the model. Corroborated across 3+ independent interview platforms.
Candidate accounts of the loop, each with its date, level, difficulty, and outcome. Scroll the feed.
2 candidate interview reports
real candidate submissions
Asked about data modeling questions. How to design different tables, how to partition data, how to scale. And asked to write SQL queries based on the previous questions. The coding interview asked simple coding questions, array, if else, loop, etc.
**Round details** Post: Senior Data Engineer Location: Bangalore Total Rounds after discussion with HR: * 1 Online Test (1 hour) * 5 Tech/Fitment Rounds (all 45 minutes) **Round Details** 1. **Online Test** * This round consisted of 4 SQL questions and a python coding question. * Total time allotted was 1 hour * All the SQL questions were regarding a holiday resort and calculations of metrics around it like People who booked the first time, Role 3 months cancellations average, busiest season, etc. Questions were between medium to hard * Python question was this: * It took me about 40-42…
Find every user active on 3 or more CONSECUTIVE days. This gaps-and-islands shape shows up in nearly every DE SQL round. Edit the query and run it against the seed data.
Problems our platform predicts for this company's interview, grouped by round. Drill the shapes that actually come up.
The patterns that sink otherwise strong candidates here.
Airbnb is a two-sided marketplace. Every metric has a host perspective and a guest perspective. Candidates who only think about the buyer side miss half the picture. When modeling bookings, consider occupancy rate (host metric) alongside conversion rate (guest metric).
Minerva is Airbnb's centralized metrics layer. It defines how every metric is calculated so that teams cannot produce conflicting numbers. If you propose a system design without mentioning metric standardization or a single source of truth, you are ignoring a core part of Airbnb's data philosophy.
Airbnb created Airflow. Surface-level knowledge (DAGs, operators, scheduling) is the minimum. Interviewers expect you to discuss backfill strategies, idempotent task design, dynamic DAG generation, SLA monitoring, and how to handle task failures gracefully in production.
Airbnb's business is highly seasonal and location-dependent. A pipeline that works for New York in July will produce very different volumes for rural Japan in February. Always mention how your design handles variable data volumes, seasonal spikes, and regional differences.
The core values interview is a real evaluation round with veto power. Candidates who prepare only for technical rounds and wing the values discussion get rejected. Prepare specific stories that demonstrate belonging, hosting, and championing the mission.
Tactical advice for the dimensions this company weighs.
Apache Airflow was born at Airbnb. Interviewers expect you to understand DAG-based orchestration, dependency management, idempotent tasks, and retry strategies. If you use Airflow, know it well. If you use a competitor (Dagster, Prefect), be ready to compare.
Airbnb trains all employees on data literacy and has a centralized metrics layer (Minerva). DEs are expected to help the entire organization, not just serve other engineers. Prepare examples of making data accessible to non-technical users.
Every Airbnb metric has a host side and a guest side. When modeling data, always consider both perspectives: a booking is revenue for the host and an experience for the guest. Interviewers notice candidates who naturally think about both sides.
Airbnb runs thousands of experiments concurrently. DE pipelines compute metrics for these experiments. Understand how to design pipelines that support A/B testing: logging experiment assignment, computing treatment vs control metrics, and handling novelty effects.
The signals behind the questions. Shape every answer around these.
Airflow was born as an internal Airbnb project in 2014 before becoming an Apache top-level project. This means Airbnb's data infrastructure runs on a deeply customized fork with proprietary operators, monitoring, and SLA tooling. The internal engineering culture treats orchestration as a first-class discipline, not just a scheduler.
Unlike most companies where metric definitions live in scattered dashboards and notebooks, Airbnb built Minerva to centralize every metric definition. DEs are responsible for registering metrics in Minerva and ensuring pipelines produce data that conforms to these definitions. This changes how you think about pipeline design: output is not just a table, it is a certified metric.
Most data engineering roles deal with one type of customer. At Airbnb, every feature, every metric, and every pipeline must account for both hosts and guests. A cancellation pipeline must update host availability, guest booking history, trust scores for both parties, and financial records. This multiplier effect makes even simple-sounding problems significantly harder.
Airbnb invests in data literacy training for all employees. Product managers, designers, and operations staff are expected to query data and interpret experiment results. DEs build infrastructure that serves the entire company, not just analysts. This means you are evaluated on how well you make data accessible, not just how well you build pipelines.
The roles behind this loop. Prep against the levels and locations they are actually filling.
Develop and automate large scale, high-performance batch and streaming data processing systems to power Airbnb’s Guest and Host products, machine learning models, and business insights.
Architect Ecosystems: Lead the design of overarching data architectures that don't just solve today’s batch needs but anticipate future real-time and AI-driven requirements.
Architect and productionize batch and real-time data systems to support various products and business needs.
The numbers, tech stack, and team structure live on the company overview.
Level-specific pages: the comp, the bar, and what the loop tests at each seniority.
Comp, level expectations, and role-specific prep.
Comp, level expectations, and role-specific prep.
Comp, level expectations, and role-specific prep.
Comp, level expectations, and role-specific prep.
Comp, level expectations, and role-specific prep.
How the role, pay, and loop stack up against peer companies.
Active recall beats re-reading by 50%
Cognitive-science meta-reviews (Dunlosky et al., 2013) rank practice testing as a top-tier study technique, while re-reading and highlighting rank near the bottom
76% of hiring managers reject on the coding task, not the resume
From HackerRank's 2024 Developer Skills Report. Candidates who look strong on paper still fail the live screen if they haven't done timed, executable practice
5 problem shapes cover 80% of data engineer loops
Dedup, sessionization, top-N-per-group, slowly-changing dimensions, partition tricks. Writing the shapes by hand turns the unfamiliar into pattern recognition