FAANG interview digest, May 10–17: freeze map, real questions, and who's paying what
This week: only Amazon and Google are actively hiring among FAANG. Hiring freeze map, Amazon SDE GenAI round breakdown, Google behavioral trap, real offers (Google L5 $550K+, Nvidia $380K, Microsoft $407K), and a 12-company TC benchmark table.
Three major tech employers pulled the hiring rug this week. Meta's ~8,000-person layoff wave begins May 20. LinkedIn cut 5% (or more, per insiders) on May 13. Netflix quietly stopped advancing candidates mid-process. Meanwhile Amazon and Google keep interviewing — slowly. If you have active loops at any of those first three companies, read Section 1 before you follow up with a recruiter.
This week's hiring heat map
The most useful thing a job seeker can do right now is sort companies into two piles before spending hours on prep work.
Still hiring:
- Amazon — Active interviewing. An SDE candidate received an offer this week after a standard 4-round process. Recruiter responsiveness is uneven (some roles are drawing only 5 applicants per opening), 1 but the process is running.
- Google — Active but genuinely slow. Team matching after a passed final loop can take 8 months. 2 Offers are being made; patience is required.
On pause or contracting:
- Netflix — Hiring freeze in effect as of May 12. 3 One Netflix employee wrote: "Sorry everyone… yeah hiring freeze, everything not already signed and accepted very likely rescinded (unless future backfill and extreme (think 0.1%) of roles)." 3 Another candidate had a final round cancelled after 11 interviews with a generic rejection email containing a blank job title field. 3 Netflix employees attribute the freeze to overhiring on the Ads team during 2026, not AI cost-cutting.
- Meta — Hiring frozen ahead of ~8,000 layoffs (~10% of global workforce) beginning May 20. 4 Zero interview reports appeared on Blind this week, consistent with a pre-layoff freeze. Six thousand open roles were also eliminated. Any active Meta pipelines are at high risk.
- LinkedIn — Executed layoffs on approximately May 13. 5 Official figure is 5% of workforce, but LinkedIn employees on Blind reported "Heard 10% in Eng" and "Half my team is gone." 5 Hiring freeze is in place.
- Microsoft — CFO Amy Hood stated on the May 13 earnings call that headcount "will decrease in coming quarters." 6 Blind users reported interview cancellations and roles closed at the final loop stage. The freeze may lift after Microsoft's fiscal year end in late June/July, per community speculation.
- Cloudflare — Cut 1,100+ employees (~21% of workforce) on May 8 in what CEO Matthew Prince and President Michelle Zatlyn described as an AI-driven restructuring toward an "agentic AI operating model." 7 The layoffs hit back-office roles; the company says it plans to hire more engineers and salespeople. Stock dropped 23% on announcement day despite record Q1 revenue.
No signal either way:
- Apple — No interview reports this week. Apple's hiring practices are notoriously opaque; the absence of data should not be read as a freeze or as active hiring.
- Salesforce, Stripe, Snowflake, Databricks, Atlassian, ServiceNow, Shopify, Workday — No interview experience reports surfaced on Blind or Reddit in the past 7 days. Not a confirmed freeze; simply no public signal.
On recruiter activity overall: one Reddit data point from a Scandinavia-based engineer (8 years of experience) noted an uptick in inbound messages over the past two months — but the companies reaching out were Home Depot, banks, and data-engineer roles, not FAANG. 8 FAANG recruiter activity remains quiet.
What they actually asked this week
Amazon
The SDE candidate who received an offer this week described a 5-stage process: 9
- Online Assessment — 2 easy-to-medium DSA problems + a Work Style Assessment
- Round 1 — 2 medium DSA problems + Amazon Leadership Principles (LP) questions
- Round 2 (System Design + "Logical and Maintainability") — Design a streaming error-log counting system with a moving average. The candidate described this round as "a bit tricky." LP questions were also included.
- Round 3 — "How have you used GenAI in your work?" + 1 medium DSA problem
- Round 4 (Hiring Manager) — All LPs: deep-dive on resume, past project decisions, what the candidate would do differently, what improvements they drove
The candidate's core advice: "Don't ignore Amazon LPs; they are just as important as DSA. Prepare 2-3 strong stories around different Amazon Leadership Principles." 9
LP stories appeared in 4 out of 5 rounds. That pattern is consistent across Amazon reports. One thing that stood out this cycle: Round 3 now has an explicit GenAI usage question — something that was not standard in earlier Amazon loops.
Two notable data points this week:
The behavioral question that got someone rejected. A Google interviewer (verified Blind profile, TC $730K) posted that a candidate answered "I'm willing to work nights and weekends" when asked "How do you handle unexpected workload?" — and was rejected. 10 The post title references "996" — the practice of working 9am to 9pm, 6 days a week, common in some Chinese tech companies. It drew 254 upvotes and 260 comments (30K views). The interviewer wrote: "The question was how they handle unexpected workload, and their answer was they willing to work nights and weekends. At least they were honest, I think. But there are smarter answers." 10
What Google wants instead: prioritization of the most important tasks, scope negotiation, delegation, stakeholder communication, and delivering value incrementally. Another Google employee in the thread shared the answer that got them hired: "I told them I would delegate, raise flags on the timeline and promptly inform all stakeholders about potential delays." 10
The 18-month process. The Google L5 UXD offer discussed in Section 3 below involved a final loop in late 2024, an offer the next morning, and then 8 months of team matching before a match was found. 2 The candidate reported that Google's fastest-growing org had recently received increased headcount budget — which is what eventually unlocked the match.
Frontier AI labs: hybrid format replacing pure LeetCode
A Blind user who spent six months interviewing at OpenAI, Anthropic, xAI, Meta's MSL/GenAI Infra team, Microsoft AI, Databricks, and several robotics and foundation-model startups compiled a detailed breakdown of what they encountered. 11 The post drew mixed responses — some commenters called it a course advertisement, others called it "genuinely useful" — but the technical specifics are consistent with known practices at these labs.
The key shift: a single loop now commonly combines coding, ML systems design, distributed systems, transformer implementation, debugging under pressure, inference serving, and evaluation design. Pure LeetCode preparation is much less effective for these roles.
What's now considered baseline knowledge for ML/AI engineer roles at frontier labs:
- Transformer implementation from scratch — multi-head attention (MHA), KV cache, Rotary Position Embedding (RoPE), Grouped Query Attention (GQA), Multi-Query Attention (MQA), causal masking, stable softmax. No
torch.nnor Hugging Face abstractions. - Numerical stability — log-sum-exp trick, float16/bfloat16 instability, NaN sources during training, layer normalization epsilon, gradient scaling
- Backpropagation derivations — softmax + cross-entropy backward pass, gradient flow through attention, why residual connections stabilize training, why pre-LayerNorm transformers converge more stably than post-LayerNorm
- Distributed training — Tensor Parallelism (TP), Pipeline Parallelism (PP), Data Parallelism (DP), ZeRO (Zero Redundancy Optimizer), FSDP (Fully Sharded Data Parallel), NCCL (NVIDIA Collective Communications Library) bottlenecks, activation checkpointing, FlashAttention, vLLM/paged attention, inference batching, GPU utilization
The candidate summarized Anthropic's focus specifically: "Hand-rolled transformer knowledge is table stakes now. Anthropic especially seemed to care a lot about: inference efficiency, tensor reasoning, memory/computation tradeoffs, cache update correctness." 11
Real offers reported this week
Google L5 UXD — $550K+ TC
An Amazon L6 UXD (Senior UX Designer, fully remote, $280K TC) accepted a Google L5 UXD offer this week after an 18-month process. 2 The candidate wrote: "I can't believe it's finally happening. It's been a year and a half I've been interviewing, team matching, etc. trying to get out of Amazon."
Offer details: $550K+ total compensation, 2 days in office. The initial offer was in the low $400Ks. The candidate leveraged a competing Meta offer to negotiate: "I was able to get an extra $177k in GSUs and a large signon that got me there." 2 (GSUs are Google Stock Units, the company's RSU equivalent.)
Level note: Amazon L6 maps to Google L5 for UX Designer roles, confirmed by the candidate's manager, ex-Amazon Googlers, their recruiter, and Levels.fyi.
The candidate's comment on the compensation gap: "The TC is a life changing jump."
Nvidia IC4 (Senior SWE) — $380K Y1 TC
An Amazon L5 engineer (8 years of experience, $210K base, hitting their 4-year RSU cliff in June with $0 unvested) shared a Nvidia IC4 (Senior Software Engineer) offer posted on approximately May 13. 12
Offer breakdown:
| Component | Amount |
|---|---|
| Base salary | $250,000 |
| RSU (4-year, 40/30/20/10 vest) | $325,000 total / ~$130K Year 1 |
| ESPP (24-month lookback, 15% discount) | Additional |
| Estimated Year 1 TC | ~$380,000 |
| 4-year average TC | ~$330,000 |
Location: Santa Clara, CA (high cost of living).
Community reaction was split. One HashiCorp employee noted that IC3s had been at $275–325K and IC5 at $500K, suggesting the IC4 floor should be closer to $400K average. An Amazon L6 commenter said they would take the offer regardless. The candidate acknowledged the slight below-average position for IC4 but accepted given the RSU cliff situation.
Microsoft L65 (Principal Software Engineering Manager) — ~$407K Y1 TC
A Blind post from approximately May 14 described a Microsoft L65 final offer. 13
Offer breakdown:
| Component | Amount |
|---|---|
| Base salary | $237,000 |
| RSU (4-year, 25/25/25/25 vest) | $320,000 total / $80,000/year |
| Signing bonus (split: joining + 1-year anniversary) | $90,000 |
| Performance bonus target | 0–30% of base |
| Estimated Year 1 TC | ~$407,000 |
Community verdict: the equity component was widely considered low for L65 level. One commenter who identified as an L64 reported receiving more in annual rewards alone than the L65's RSU grant.
Compensation benchmarks: senior-level medians
| Company | Level (Senior SWE equiv.) | Base | Stock/yr | Bonus | Total Comp | Vest schedule |
|---|---|---|---|---|---|---|
| Databricks | L5 | $213K | $415K | $18.7K | $647K | 40/30/20/10 (front-loaded) |
| Snowflake | IC3 | $244K | $295K | $26.4K | $565K | 25/25/25/25 |
| Netflix | L5 | $526K | $0 | $0 | $526K | All cash, no RSUs |
| Meta | E5 | $225K | $219K | $28.7K | $473K | 25/25/25/25 |
| Apple | ICT5 (Staff) | $260K | $171K | $27.4K | $458K | 25/25/25/25 |
| Stripe | L3 | $233K | $181K | $23.3K | $437K | 25/25/25/25 |
| L5 | $224K | $168K | $37.3K | $430K | 38/32/20/10 (front-loaded) | |
| Amazon | L6 | $218K | $183K | $1.1K | $402K | 5/15/40/40 (backloaded) |
| Apple | ICT4 (Senior) | $216K | $94.7K | $22.7K | $333K | 25/25/25/25 |
| IC3 | $208K | $83K | $18.7K | $310K | 25/25/25/25 | |
| Salesforce | Senior MTS | $201K | $37.9K | $24.1K | $263K | 25/25/25/25 |
| Microsoft | Level 62 | $174K | $32K | $14.3K | $220K | 25/25/25/25 |
Three things the table doesn't show:
Amazon's vesting schedule is not like others. The 5/15/40/40 schedule means only 5% of the total RSU grant vests in Year 1. On an L6 grant with $183K/year average stock value (i.e., ~$732K total over 4 years), that's roughly $37K in stock Year 1 — versus the ~$183K implied by the average. Amazon offsets this with sign-on bonuses paid in years 1 and 2, but actual Year 1 cash is materially lower than the $402K stated average. Verify the sign-on terms specifically before accepting any Amazon offer.
Databricks L4 (mid-level) already clears FAANG senior TC at most companies. Databricks L4 total compensation is approximately $411K ($175K base + $220K stock/yr + $16.5K bonus), 22 which exceeds the senior-level median at Google ($430K is close), Microsoft ($220K), Apple ICT4 ($333K), LinkedIn ($310K), and Salesforce ($263K). The stock at Databricks ($415K/yr at L5) is nearly twice the base salary — evaluate the current stock price and growth trajectory before assuming those paper numbers hold.
Netflix pays almost entirely in cash. L5 TC of $526K is all base salary — no RSUs, no annual bonus at L4+. 18 This is deliberate: Netflix's philosophy is that engineers should be trusted to manage their own financial exposure to company risk. Given this week's hiring freeze, that full-cash offer may not be accessible until the freeze lifts.
How to use this heading into next week
If you're interviewing at Amazon: Treat LP preparation as equal in weight to DSA. Plan 2-3 stories mapped to different LPs, and be ready for a dedicated GenAI usage discussion in Round 3. The system design question this week was streaming error-log counting with moving average — worth adding to your prep list.
If you're in Google's pipeline: Accept that team matching can take months after your offer. Do not turn down other offers while waiting unless you have signed Google paperwork. For behavioral rounds, build answers around prioritization and delegation frameworks, not availability. Using a competing offer as leverage worked this week — a Meta offer added $177K in GSUs to a Google package. 2
If you're targeting frontier AI labs: Build your transformer knowledge from scratch. The hybrid interview format — combining systems thinking, distributed training knowledge, and live debugging — is not addressable by grinding LeetCode alone. Anthropic specifically probes inference efficiency and tensor reasoning.
If you have active loops at Netflix, Meta, Microsoft, or LinkedIn: Treat any loop as at risk. Ask your recruiter directly whether the role is a backfill or new headcount, and whether the position is still actively open. Do not stop interviewing elsewhere while waiting on any of these four companies.
On negotiation: Every real offer posted this week involved a candidate with leverage — either a competing offer, a cliff expiring, or a market reference point. The Levels.fyi benchmarks above are the baseline; your ability to negotiate above them depends on the alternatives you can credibly present.
Cover image from: A Meta employee gets real about the horror of working there right now
References
- 1Blind: Lack of Talent at Amazon
teamblind.com
- 2Blind: Goodbye Amazon, Hey Google!
teamblind.com
- 3Blind: Netflix L5 — No recruiter response
teamblind.com
- 4
- 5
- 6
- 7
- 8
- 9Blind: Amazon SDE Interview Experience
teamblind.com
- 10
- 11
- 12Blind: Nvidia IC4 offer
teamblind.com
- 13Blind: L65 offer at Microsoft
teamblind.com
- 14
- 15
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.