The Waymo SWE Interview Guide
See how Waymo runs its software engineer loop, from the coding phone screen to the machine-learning and system-design rounds that decide most outcomes. Then drill the real questions candidates report, so the autonomy-scale problems in the room feel familiar rather than foreign.
- Loop length: 4 to 6 weeks
- Expected difficulty: Medium to hard
- Core language: C++ heavy
- Onsite weight: ML and system design
Waymo levels & pay
- SWE II (L3, Entry)
- SWE III (L4, Mid)
- Senior SWE (L5, Senior)
- Staff SWE (L6, Staff)
- Senior Staff (L7, Sr Staff)
- Principal (L8, Principal)
The Waymo SWE interview, round by round
Recruiter screen (30 min · call)
The opening conversation. Your recruiter confirms your background, the level and team you are targeting, your timeline, and comp expectations, and gives you a first read on which part of the stack you would join, perception, planning, simulation, or infrastructure. It is rarely technical, but it decides your level and which specialist rounds you will see.
- Signal your stack. Perception and planning teams weight machine learning and C++ far more heavily than infrastructure roles, so say where your strength lies.
- Anchor your level early. Senior versus staff changes the difficulty of the whole loop and how much the design and ML rounds count.
- Know your comp range. Waymo pays on Alphabet's equity structure, so understand how GSU grants and refreshers work before you name a number.
Coding phone screen (45 min · shared editor)
One medium coding problem with a Waymo engineer on a shared editor, most often expected in C++. This is where communication counts as much as the answer: clarifying constraints, narrating your approach, and reasoning through edge cases out loud is what separates a clean pass from a borderline call. Expect follow-ups that push on complexity and memory.
- Write real C++. Interviewers watch for idiomatic memory handling and correct use of the standard library, not pseudocode that happens to run.
- Reach an optimal solution and explain the trade-offs as you build it, rather than coding in silence and revealing the plan at the end.
- Edge cases carry weight. Autonomy code fails on the rare case, so interviewers reward candidates who surface boundary conditions unprompted.
Onsite coding (2 rounds · 45 min each)
Two coding rounds on the virtual onsite, harder than the phone screen and still typically in C++. Problems lean toward data structures, graphs, geometry, and the kind of state management that shows up in robotics and real-time systems. Some teams add a debugging or code-reading exercise over an existing snippet rather than a blank editor.
- Expect geometry and graphs. Grid traversal, spatial reasoning, and pathfinding recur because they mirror real autonomy problems.
- Performance is part of correctness. A solution that is technically right but allocates on a hot path will draw follow-ups about latency.
- Read code as carefully as you write it. Some rounds hand you existing code to extend or fix, which rewards careful tracing over speed.
ML & system design (45 to 60 min · whiteboard)
The round that decides most senior loops. Depending on the team, you either design a machine-learning system or a large software system for autonomy. Common prompts ask you to design how a self-driving model is evaluated end to end, or how behavior prediction is built from sensor and camera data. Interviewers push on metrics, data pipelines, failure modes, and how you would know the system is safe.
- Design the evaluation, not just the model. How you measure a driving policy end to end, including rare-event and safety metrics, matters more than the architecture diagram.
- Reason about sensor and camera data. Behavior-prediction prompts expect you to handle multi-modal input, uncertainty, and time, not a single clean feature vector.
- Name the failure modes. Strong candidates talk about edge cases, distribution shift, and how the system degrades safely, not only the happy path.
Behavioral & team match (45 min · call)
A behavioral round paired with team-fit conversation, sometimes with a hiring manager. Waymo probes ownership, how you handle ambiguity in a research-heavy environment, and how you make decisions when safety and shipping pull against each other. The signal from this round feeds the hiring committee alongside your technical scores before an offer and team are confirmed.
- Bring safety-versus-speed stories. Waymo values engineers who can show judgement when correctness cannot be compromised.
- Show comfort with ambiguity. Much of the work is unsolved, so interviewers look for people who move forward without a clean spec.
- Team match is real. Your fit with a specific team, perception, planning, simulation, or infra, is decided here as much as your general strength.
How hard is the Waymo interview?
Difficulty mix: 15% easy, 70% medium, 15% hard.
How to prepare
Build a daily C++ coding rhythm (Weeks 1 to 5)
Code every day, in C++ if the language is optional for you, since most rounds expect it. Rotate through the patterns Waymo leans on, graphs and BFS or DFS, grids and geometry, hash maps, heaps, and careful memory handling, starting at medium and climbing into hard, always timed on a shared editor.
Refresh machine-learning and autonomy fundamentals (Weeks 1 to 3)
Review the ML concepts that surface for perception and planning teams: how models are trained and evaluated, metrics for imbalanced and rare events, and how sensor and camera data are represented. Even infrastructure candidates benefit from being able to talk about the autonomy stack at a high level.
Drill ML and system design for autonomy (Weeks 2 to 4)
Once your coding is warm, work through full designs end to end: how you would evaluate a self-driving model, build behavior prediction from sensor data, or scale a simulation and data pipeline. Say each decision aloud, cover metrics, data, and failure modes, and prepare for follow-ups on safety and edge cases.
Run full mock loops with behavioral reps (Week 5)
In the last week, put it together: sit a few complete loops back to back, coding plus an ML or design round plus behavioral, in single sittings. Draft ownership and safety-versus-speed stories in STAR form early so they are rehearsed, and practice holding focus across four or five interviews in a row.
Recently asked
Implement a rate limiter
About Waymo
Waymo builds the Waymo Driver, an autonomous driving system that operates a public robotaxi service across several US cities. It began in 2009 as the Google self-driving car project and became an independent company under Alphabet in 2016. Its engineering spans perception, behavior prediction, planning, simulation, mapping, and the large-scale infrastructure that trains and validates the driver on data from a growing fleet.
Because the product is a physical system operating on public roads, the bar centers on safety, rigorous evaluation, and machine learning at scale rather than pure feature velocity. Much of the work sits at the boundary of research and production, where models must be measured against rare, high-stakes events and code correctness is treated as a safety property, not just a quality one.
Frequently asked
How many rounds is the Waymo interview?
Typically five stages: a recruiter screen, a coding phone screen, and a virtual onsite of coding rounds plus a machine-learning or system-design round and a behavioral round. Team match and a hiring committee review sit at the end before an offer.
How hard is the Waymo SWE interview?
Medium to hard. The coding bar is beatable with focused pattern practice, but the machine-learning and system-design rounds are where senior loops are usually decided, and they expect real depth on autonomy and evaluation.
How important is C++ at Waymo?
Very. Most of the autonomy stack is written in C++, and coding rounds usually expect it. If C++ is not your strongest language, invest early in idiomatic memory handling and standard-library fluency before the loop.
Do I need machine-learning knowledge to interview at Waymo?
For perception, planning, and prediction teams, yes. You should be able to reason about training, evaluation metrics, imbalanced data, and sensor and camera inputs. Infrastructure and platform roles weight ML less, but a high-level grasp of the stack still helps.
How much robotics background do I need?
You do not need a robotics degree, but familiarity with geometry, spatial reasoning, state estimation, and real-time constraints helps, since coding problems often mirror those domains. Strong general engineers regularly convert without a formal robotics background.
What does the system-design round cover?
Often something autonomy-specific: how you would evaluate a self-driving model end to end, design behavior prediction from sensor data, or scale a simulation and data pipeline. Interviewers push on metrics, failure modes, and how you know the system is safe.
Is the ML round the same as system design?
It depends on the team. Some candidates get a machine-learning system design, others a classic large-system design. Either way the round tests how you structure, measure, and de-risk a component of the autonomy stack, not just the diagram.
What coding topics show up most?
Graphs and traversal, grids and geometry, hash maps, heaps, and problems with careful state management and memory. Because the work is real-time and safety-critical, edge cases and performance on hot paths are weighted heavily.
Does Waymo ask behavioral questions?
Yes. A dedicated behavioral round probes ownership, handling ambiguity in a research-heavy setting, and judgement when safety and shipping conflict. Prepare STAR stories, including at least one where correctness could not be compromised.
How does Waymo pay compare, and how does the equity work?
Waymo is an Alphabet subsidiary and pays on Alphabet's compensation structure, with equity granted as Google Stock Units (GSUs) that vest over four years, plus periodic refreshers. Total compensation is competitive with Google for the equivalent level.
What are the levels at Waymo?
The ladder aligns with Alphabet's, roughly L3 for entry engineers up through L5 senior, L6 staff, and L7 and L8 for senior staff and principal roles. Your level is set early and shapes how heavily the design and ML rounds count.
How long does the Waymo interview process take?
Usually four to six weeks from the recruiter screen to an offer, though scheduling the onsite and team matching can extend it. Specialist rounds and hiring-committee review add time compared with a lighter loop.
Where does Waymo hire engineers?
The main hub is Mountain View, with engineering also in San Francisco, Phoenix, Austin, and additional sites tied to operations and hardware. Most software roles concentrate in the Bay Area.
How should I prioritize my prep?
Keep a daily C++ coding rhythm, refresh machine-learning and autonomy fundamentals, then drill ML and system design for autonomy scenarios like model evaluation and behavior prediction. Layer in behavioral stories last so they are rehearsed.
Is domain knowledge of self-driving required?
Not strictly, but it helps a lot. Being able to talk about perception, prediction, planning, and how a driving system is evaluated and kept safe signals genuine interest and lets you go deeper in the design and ML rounds.
Waymo interview questions on Karavine