Interview Guide

Amazon Data Engineer Interview in San Francisco Bay Area (L5)

Hiring for Data Engineer at Amazon (L5) runs Leadership Principles woven into every round, with a Bar Raiser holding veto power. The hiring bar is shipped production pipelines end-to-end and can debug them when they break; the median candidate brings 2-5 years of DE experience. Below we dig into how this runs out of the San Francisco Bay Area office (San Francisco / South Bay, CA), with cost-of-living-adjusted compensation.

Compensation

$155K–$185K base • $230K–$290K total (L5)

Loop duration

3.8 hours onsite

Rounds

5 rounds

Location

San Francisco / South Bay, CA

Compensation

Amazon Data Engineer in San Francisco Bay Area total comp

Across 139 samples

Offer-report aggregate, 2018-2026. Level mapped: L5. Typical experience: 4-9 years (median 6).

25th percentile

$220K

Median total comp

$280K

75th percentile

$316K

Median base salary

$180K

Median annual equity

$75K

Median total comp by year

2021
$288K n=31
2022
$281K n=28
2023
$227K n=20
2024
$264K n=19
2025
$281K n=16
2026
$297K n=24

5 currently open data engineer postings in San Francisco Bay Area.

Tech stack

What Amazon data engineers actually use

Across 5 open roles

These are the tools that show up in Amazon's DE job descriptions right now in San Francisco Bay Area. Click any chip to drop into an interview prep page for it.

Cloud
AWS
Infra
Kubernetes
Updated from current job listings
Prepare for the interview
01 / Open invite
02min.

Walk into Amazon knowing the Python pattern they'll test.

a Amazon Python query, the same shape a screen would give you.
The diff against expected. Where ties broke. What you missed.
sandbox
1def sessionize(events):
2 sessions = []
3 for e in events:
4 if gap_minutes(e) > 30:
5
Execute your solution0.4s avg.
AmazonInterview question
Solve a Amazon problem

Round focus

Domain concentration by round

Across 5 job descriptions

Where each domain tends to come up in Amazon's loop, derived from 5 current data engineer job descriptions. Longer bars mean heavier weight.

Online Assessment

Python88%
SQL47%
Architecture10%
Spark8%
Modeling6%

Phone Screen

SQL71%
Python68%
Architecture24%
Spark12%
Modeling8%

Onsite Loop

Architecture59%
Modeling36%
Python31%
SQL30%
Spark13%

Practice problems

Amazon data engineer practice set

4 problems

Interview problems predicted for Amazon data engineers based on their actual job descriptions. Click any problem to work it in a live coding environment.

SQLmedium~10 min

The Holdouts

In our push-notification log, each message records the plan tier it targeted in the `platform` field. Find the unique users who were sent a `basic`-tier notification but never a `premium`-tier one.

Open in practice environment
Pythonhard~5 min

The Overlap

Your monitoring system logs server maintenance as `[start, end]` minute ranges, and windows that overlap or sit back-to-back really describe one continuous outage. Collapse the `windows` so any that overlap or touch at an endpoint become a single range, and return them ordered by start time. Two windows touch when one ends exactly where the next begins.

Open in practice environment
Modelingmedium~20 min

Split Decision

We run dozens of product experiments at once on our consumer app, and each user who enters an experiment is locked to a single variant for its duration: either the control or one of the treatment groups. Analysts measure lift by comparing what users in each variant did after they were enrolled, joining their assignments back to our existing event log. Design the data model so a user can never land in two variants of the same experiment and so only post-enrollment behavior counts toward a variant's metrics.

Open in practice environment
SQLeasy~10 min

The Upper Rungs

A compensation team is setting reference points for a new salary-band ladder, where a value that appears more than once still counts only once. Return the five highest values from the employee metrics table, highest first.

Open in practice environment

Top 2 sellers by revenue in each marketplace

Classic DE round opener. Window function + partition. Edit to tweak the threshold.

WITH seller_totals AS (
SELECT
marketplace,
seller_id,
SUM(amount) AS revenue
FROM seller_orders
GROUP BY marketplace, seller_id
),
ranked AS (
SELECT
marketplace,
seller_id,
revenue,
DENSE_RANK() OVER (
PARTITION BY marketplace
ORDER BY revenue DESC
) AS rk
FROM seller_totals
)
SELECT
marketplace,
seller_id,
revenue
FROM ranked
WHERE rk <= 2
ORDER BY marketplace, revenue DESC

San Francisco / South Bay, CA

Amazon in San Francisco Bay Area

The reference market for US tech comp. Highest base DE salaries in the US, highest cost of living, deepest senior-engineer hiring pool.

Amazon's San Francisco Bay Area office hires at the company's reference compensation band. The San Francisco Bay Area office's interview loop mirrors the global loop structure; team assignment and comp-band negotiation are the main local variables.

Prepare for the interview
03 / From the bank03 of many
03hand-picked.

The Safe Caster

Easy8 min

Type conversion is easy, until it is not.

Pulled from debriefs where Python parsing was the gate.

The loop

How the interview actually runs

01Recruiter screen

30 min

Logistics, team fit, and a light Leadership Principle question. Recruiters confirm seniority expectations before booking the loop. Misalignment here can downlevel the loop.

  • Have a 60-second pitch that names 2-3 concrete data systems you've built
  • Confirm the team. Amazon has hundreds of DE teams across AWS, Retail, Ads, Alexa, Prime Video, Pharmacy
  • Ask about the comp band early to avoid end-of-loop misalignment

02Technical phone screen

60 min

1 SQL problem, 1 Python or pipeline design problem, and 10-15 min of Leadership Principle questions. The SQL is harder than the Online Assessment, expect multi-step window functions or self-joins.

  • Narrate approach before writing code. Amazon grades process, not just the final answer
  • Name the LP before telling the story
  • Prepare at least 2 stories per LP; follow-ups probe a third story on the same theme

03Onsite: SQL deep-dive

60 min

2 to 3 SQL problems with escalating difficulty, usually in Amazon contexts (seller performance, order fulfillment, inventory). Ends with 10 min of LP questions.

  • Practice window functions across large partition cardinalities
  • Be ready to rewrite correlated subqueries as joins and vice versa
  • When asked about optimization, mention partition pruning and columnar storage

04Onsite: Bar Raiser

60 min

An interviewer from outside the hiring team with veto power. Heaviest on Leadership Principles, with one harder technical problem. Checks whether you raise Amazon's hiring bar.

  • Bring a story where you were wrong and had to change course
  • Quantify impact: cost saved, latency reduced, users affected
  • If you don't know something, say so. Fabricating kills the loop faster than any technical gap

Level bar

What Amazon expects at L5 Data Engineer

Pipeline ownership

Mid-level DEs own pipelines end-to-end. Interviewers expect stories about designing, deploying, and maintaining a data pipeline that has been in production for 6+ months.

SQL + Python or Spark fluency

SQL is the floor. Most teams also expect fluency in either Python for data manipulation (pandas, airflow DAGs) or Spark for larger-scale processing.

On-call debugging

You should have concrete stories about production incidents: what alert fired, how you diagnosed, what you fixed, and what post-mortem action you owned.

Amazon-specific emphasis

Amazon's loop is characterized by: Leadership Principles woven into every round, with a Bar Raiser holding veto power. Calibrate your preparation to that, generic FAANG prep will not close the gap on company-specific expectations.

Behavioral

How Amazon frames behavioral rounds

Dive Deep

The most relevant LP for data engineers. Amazon wants DEs who trace anomalies through 3+ layers of the stack instead of patching symptoms.

Tell me about a time you found a data quality issue that others had missed.

Ownership

You built it, you own it, including on-call and long-term maintenance. Ownership extends beyond your explicit scope when dependencies break.

Describe a situation where you went beyond your role to solve a problem.

Bias for Action

Speed beats perfection. Amazon wants DEs who ship V1 in 2 weeks rather than a perfect solution in 3 months.

Tell me about a time you made a decision without having all the information.

Earn Trust

Trust comes from delivery and transparency. Bar Raisers check whether you can admit mistakes and communicate setbacks without spinning.

Tell me about a time you delivered bad news to a stakeholder.

Prep timeline

Week-by-week preparation plan

8-10 weeks out
01

Foundations and gap analysis

  • ·Do 10 medium SQL problems. Note which patterns feel slow
  • ·Write out 2-3 behavioral stories per value, Amazon weights this round heavily
  • ·Read Amazon's public engineering blog for recent architecture patterns
  • ·Review your prior production work, pick 3-5 projects you can discuss in depth
6 weeks out
02

SQL and coding fluency

  • ·Practice window functions until DENSE_RANK, ROW_NUMBER, LAG, LEAD are reflex
  • ·Do 20+ Amazon-style problems in their domain
  • ·Time yourself: 25 min per medium, 35 min per hard
  • ·Record yourself narrating your approach aloud. Interviewers weigh how you explain it, not only what you write
4 weeks out
03

Pipeline awareness and behavioral depth

  • ·Review pipeline architecture basics: idempotency, partitioning, backfill
  • ·Practice explaining a pipeline you've worked on end-to-end in 5 minutes
  • ·Refine behavioral stories based on mock feedback
  • ·Do 10 more SQL problems at medium difficulty
2 weeks out
04

Behavioral polish and mock loops

  • ·Rehearse every story out loud. Cut to 2-3 minutes each
  • ·Run 2 full mock loops with a mid-level DE or coach
  • ·Identify your 3 weakest behavioral areas and draft additional stories
  • ·Review recent Amazon news or earnings call for fresh talking points
Week of
05

Taper and logistics

  • ·No new content. Review your notes only
  • ·Sleep. Mental energy matters more than one more practice problem
  • ·Confirm logistics: laptop charged, shared-doc tool tested, snack and water nearby
  • ·Remember: interviewers want to find reasons to hire you, not to reject you

FAQ

Common questions

What level is Data Engineer at Amazon?
At Amazon, Data Engineer corresponds to the L5 level. The bar emphasizes shipped production pipelines end-to-end and can debug them when they break without people-management responsibilities.
How much does a Amazon Data Engineer in San Francisco Bay Area make?
Looking at 139 sampled offers from 2018-2026, Amazon Data Engineer in San Francisco Bay Area total comp comes in at $280K median, ranging from $220K to $316K, median base $180K and median annual equity $75K. Typical experience range: 4-9 years..
Does Amazon actually hire data engineers in San Francisco Bay Area?
Yes, Amazon maintains a San Francisco Bay Area office and hires Data Engineer data engineers there. Team assignment may be office-locked or global; confirm with the recruiter before the loop.
How is the Data Engineer loop different from other levels at Amazon?
The format of the loop matches other levels; difficulty and evaluation shift to shipped production pipelines end-to-end and can debug them when they break, and questions at this level dig into production pipeline ownership and on-call debugging.
How long should I prepare for the Amazon Data Engineer interview?
Most working DEs find 6-8 weeks is about right. The technical prep scales with experience; the behavioral story bank is where candidates underestimate time.
Does Amazon interview data engineers differently than software engineers?
Yes, the DE track at Amazon emphasizes SQL depth, warehouse and pipeline design, and real production data experience (late data, backfills, quality checks), which generalist SWE loops don't test.