Google · Phone screen → Onsite
In processGoogle FDE and Cloud AI, Experienced: 6 Experiences (Part 4 of 4)
Google Forward Deployed Engineer loops for experienced hires in 2026 paired a coding round with a GenAI design or scenario round; one Cloud AI loop too.
At a glance
- Process: FDE: one coding round plus one design, scenario or role-related knowledge (RRK) round (5 of 5). Cloud AI Engineer: coding, two RRK design rounds, then Googleyness (1 report).
- Asked recently: GenAI or agent design or scenarios in 4 of 5 FDE reports, such as a CTO role-play for an AI-powered system (FDE 03); evaluate a prefix expression (FDE 04); a binomial sample-size statistics RRK (FDE 05).
- Difficulty: FDE coding was LeetCode medium to medium-hard, with no AI tools allowed (FDE 03); two candidates chose an agent or workflow design track over media generation (FDE 01, 03).
- Practise first: enterprise AI agent design, evaluate a prefix expression and binomial sample size.
Coverage: Experienced hires. Forward Deployed Engineer: 5 reports, May to September 2026 (FDE III or FDE, GenAI; Europe and India named). Cloud AI Engineer: one Conversational AI Engineer report from Paris, March 2024. Forward Deployed Engineer is a customer-facing Google Cloud role, separate from SWE III, so the two roles are counted apart.
In this series
- Part 1: L4 ML-track loops from fall 2026 back to January 2026, plus the topic summary for all 28 ML-track reports.
- Part 2: ML-track loops from November 2025 back to June 2025.
- Part 3: ML-track loops from May 2025 back to March 2024, then patterns across the ML track.
- Part 4 (current): Forward Deployed Engineer and Cloud AI Engineer loops, each in its own role section.
FDE and Cloud AI reports from candidates who did not state their experience: Google FDE and Cloud AI, hiring level not reported.
Forward Deployed Engineer (5 reports)
What came up
All five FDE reports describe one coding round and one design, scenario or role-related knowledge (RRK) round. Four of the five describe a GenAI or agent-focused design or scenario round.
| Topic | Reports | Last seen |
|---|---|---|
| Coding round | 5 of 5 | Sep 2026 |
| GenAI or agentic design or scenario round | 4 of 5 | Sep 2026 |
| A choice of design track | 2 of 5 | Sep 2026 |
Expression evaluation (FDE 04) and statistics (FDE 05) came up once each. A candidate can sit in several rows.
Questions we found
Each question comes from one candidate.
GenAI / AI System Design
| Question | Reports | Last seen |
|---|---|---|
| Role-play: a CTO wants an AI-powered system (agent-design track); filed under design an enterprise AI agent over multiple data sources | 1 | Jul 2026 (report date) |
Scenario Questions
| Question | Reports | Last seen |
|---|---|---|
| Situational scenarios about agentic AI workflows | 1 | Aug 2026 (report date) |
Statistics & Probability
| Question | Reports | Last seen |
|---|---|---|
| Sample size that keeps a confidence-interval half-width under 2% | 1 | May 2026 (report date) |
Algorithms / Data Structures
| Question | Reports | Last seen |
|---|---|---|
| Prefix-expression evaluation-style problem (evaluate a prefix expression) | 1 | May 2026 |
5 detailed candidate timelines
RRK is the role-related knowledge round.
FDE 01: FDE III loop in Europe with a coding round and a design round
When: GenAI design between 2026-08-28 and 2026-09-07 · Level: FDE III; the L-number isn't stated · Where: Europe · Outcome: offer not reported as of 2026-09-07
- Coding: a LeetCode medium, not named.
- GenAI system design: the candidate planned, ahead of the round, to take the workflow / enterprise track over image and video generation; prompt not shared.
FDE 02: FDE III (L5) GenAI loop with a scenario assessment and an offer
When: report 2026-08-02 (year inferred) · Level: L5, as written · Outcome: offer 2026-08-06 (year inferred)
- Coding: one LeetCode problem, not named.
- Assessment: situational scenarios about agentic AI workflows.
A director review followed the interviews; there was no hiring-manager round.
FDE 03: FDE L4 loop with a CTO role-play design, rejected
When: report 2026-07-07 · Level: L4, as reported · Outcome: rejected by email, after 2026-07-08
- System architecture, role-play (60 min): a CTO wants an AI-powered system for the company; agent-design track picked over audio/video; topics: architecture and end-to-end design, scalability, security, evaluation, deployment, and scoping, requirements and discovery. FastPrep files it under design an enterprise AI agent over multiple data sources.
- Coding (60 min): a single open-ended problem set in a business scenario, about LeetCode medium; no AI tools allowed; the problem itself wasn't shared.
FDE 04: Google FDE loop with a prefix-expression coding round
When: 2026-05-26 · Level: L4–L5 band; the candidate aimed for L5 · Outcome: not reported as of 2026-05-27
- Coding: a prefix-expression evaluation-style problem, difficulty not stated; practice: evaluate a prefix expression.
- RRK (system-design style): moved to the following week; whether it was held isn't reported.
FDE 05: Google FDE GenAI loop in India with a statistics RRK, rejected
When: report 2026-05-20 (year inferred) · Level: not stated · Where: India · Outcome: rejected
- Coding: one LeetCode medium-hard problem, not named.
- RRK: a statistics problem, find the sample size N that keeps a confidence-interval half-width under 2%, using the CLT and the binomial; practice: sample size for a binomial margin of error. A terse later reply seems to liken the RRK's depth to expert-level GenAI system design.
Patterns across the FDE reports
-
Coding plus design. All five describe one coding round and one design, scenario or RRK round. FDE 04's RRK was moved a week, and whether it was held isn't reported.
-
A choice of design track. FDE 01 planned to take the workflow / enterprise track over image and video generation, and FDE 03 took agent design over audio/video systems. In both, an agent or workflow option sat next to an audio, image or video option.
Cloud AI Engineer (1 report)
What came up
The one Cloud AI Engineer report is a 2024 Conversational AI loop: a coding round with two medium two-pointer problems, two RRK rounds on conversational-AI design, and a final Googleyness round.
Questions we found
GenAI / AI System Design
| Question | Reports | Last seen |
|---|---|---|
| End-to-end chatbot / virtual-agent design (design a live-session conversational chatbot) | 1 | Feb 2024 |
1 detailed candidate timeline
Cloud AI 01: Conversational AI Engineer loop in Paris with two RRK design rounds
When: final round ~2024-03-04 (inferred) · Level: L4 · Where: Paris · Outcome: not reported
- Coding: two medium two-pointer problems, not named.
- RRK 1 and RRK 2 (by 2024-02-26): two conversational-AI system design rounds on chatbots and virtual agents end to end; the brief for both covered design, monitoring, performance and training; practice: design a live-session conversational chatbot.
- Googleyness (~2024-03-04): the final behavioral round.
Common questions
What is the Google FDE interview process for experienced hires?
One coding round and one design, scenario or RRK round in all five reports (FDE 01–05); one offer came after a director review with no hiring-manager round (FDE 02).
What does the Google FDE design round ask?
GenAI and agent systems: a CTO role-play for an AI-powered system (FDE 03), agentic workflow scenarios (FDE 02), and a GenAI design on a chosen track (FDE 01).
Is the Google FDE coding round hard?
LeetCode medium in FDE 01 and FDE 03, medium-hard in FDE 05; FDE 03's was an open-ended business scenario with no AI tools allowed.
Keep practicing
- Design: enterprise AI agent over multiple data sources (FDE 03) and live-session conversational chatbot (Cloud AI 01).
- Coding and statistics: evaluate a prefix expression (FDE 04) and binomial sample size for a margin (FDE 05).
- Google question banks: system design and algorithms (sign-in needed).
- Everything Google in one place: the Google interview page and Google hiring insights.
- More firsthand reports: the Experience feed.
In this series
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.