Invent and Simplify (Amazon Leadership Principle)
Interviewers want proof you removed complexity, not just added a clever feature. The 'Simplify' half is what most candidates miss.
About this theme
What interviewers are evaluating
- →Did you invent a genuinely new approach, or just apply a standard pattern and call it innovation?
- →Did you simplify - removing steps, services, or concepts - or only add new complexity on top?
- →Were you externally aware - did you borrow ideas from other teams, papers, or open source rather than reinventing?
- →Did the simplification reduce cost or burden for someone downstream (callers, on-call, future maintainers)?
- →Did you resist over-engineering - choosing the simplest thing that solves the real problem?
- →Can you quantify the complexity you removed (lines of code, services retired, steps eliminated, latency cut)?
Common prompts
Variations on these are asked at every level. Have a story pre-loaded for at least three of them.
- ?Tell me about your most innovative idea or project.
- ?Describe a time you simplified a complex process or system.
- ?Tell me about a time you invented something new to solve a problem.
- ?Give an example of when you found a simpler solution to a problem others were overcomplicating.
- ?Tell me about a time you challenged the status quo with a new approach.
- ?Describe a time you eliminated a process or piece of technology that was no longer needed.
- ?Tell me about a time your simple solution beat a more complex proposal.
Sample STAR answers
Both strong and weak examples, with notes on what makes each work (or fail). Read the weak examples carefully - the patterns they show up are the ones interviewers are trained to spot.
Strong: Replaced a 4-service pipeline with one queue consumer
- Situation
- Our notification system had grown into a chain of four microservices and two SQS queues that turned an event into an email. It had been built incrementally over two years, and each hop added retries and its own config. We had three production incidents in a quarter where a notification silently died between hops, and on-call spent an average of 90 minutes per incident tracing where the message dropped.
- Task
- I was the senior engineer on the platform team. Rather than add yet another monitoring layer on top of the chain, which was the team's first instinct, I argued we should ask whether the chain needed to exist at all.
- Action
- I diagrammed the actual data flow and found that three of the four services only enriched the payload with data that was already available at the source event. I built a proof of concept that collapsed the pipeline into a single consumer that read the raw event, enriched inline, and sent the email, with one idempotency key end to end. I borrowed the idempotency-key pattern from a payments team in the company rather than inventing my own. I ran it in shadow mode against production traffic for a week to prove parity before cutting over, then migrated one notification type at a time.
- Result
- We retired two services and one queue. Median end-to-end latency dropped from 4.2 seconds to 600 milliseconds. The silent-drop incident class went to zero over the following two quarters, and on-call pages for notifications fell by roughly 70 percent. We deleted about 6,000 lines of code.
What makes this strong: (1) The 'simplify' half is the headline - services retired, code deleted. (2) Externally aware - borrowed a proven pattern instead of inventing one. (3) De-risked with shadow mode before cutover. (4) Quantified complexity removed and downstream burden lifted (on-call pages).
Strong: Invented a self-serve config to kill recurring tickets
- Situation
- I owned an internal experimentation platform. Every time a product team wanted a new feature flag, they filed a ticket and waited 2 to 3 days for my team to hand-edit a config file and deploy. We were getting about 40 such tickets a month and they were the top source of interrupt work, pulling roughly a day a week off the team.
- Task
- I wanted to eliminate the team as the bottleneck entirely, not just process the tickets faster.
- Action
- Instead of building a heavy admin UI, which would have been months of work, I invented a lightweight approach: a schema-validated YAML file in each team's own repo that our platform watched and hot-loaded, with a CI check that rejected malformed flags before merge. Product teams now owned their own flags through a pull request, the workflow they already lived in. I kept the surface tiny on purpose - I resisted requests to add targeting rules in v1 because that would have reintroduced the complexity I was trying to remove. I wrote a one-page doc and ran two office-hours sessions to onboard teams.
- Result
- Flag-creation tickets dropped from about 40 a month to under 3 within two months, and median time-to-create went from 2 to 3 days to under 10 minutes. My team got back roughly a day a week. Three other internal platform teams adopted the same repo-watching pattern after I shared it.
What makes this strong: (1) Genuine invention but the simplest viable version - chose YAML-in-repo over a heavyweight UI. (2) Actively resisted scope creep to avoid re-adding complexity. (3) The result removed the candidate's own team as a bottleneck. (4) The pattern spread, showing it was a real simplification others valued.
Weak: Vague 'I built a cool new tool'
- Situation
- Our team had an old system that was kind of slow and hard to use.
- Task
- I thought we could do something more modern and innovative.
- Action
- I built a new dashboard using the latest framework and added a bunch of new features people had been asking for. It used some cutting-edge tech that nobody else on the team had tried.
- Result
- People liked the new dashboard and it looked a lot better. It was a more innovative solution than what we had before.
Why this is weak: (1) No simplification at all - it only added features and 'cutting-edge tech.' (2) Adopting a new framework nobody else knew increases complexity and bus-factor risk. (3) No metrics - 'slow' and 'people liked it' are not measurements. (4) 'Innovative' is asserted, not demonstrated; there's no real problem being solved.
Common pitfalls
- ×Telling an 'invention' story that only added complexity - a clever layer nobody else can maintain is the opposite of this principle.
- ×Ignoring the 'Simplify' half entirely and only talking about new features you shipped.
- ×Adopting trendy technology for its own sake instead of because it was the simplest fit for the problem.
- ×No measurement of the complexity removed (services retired, code deleted, steps eliminated, latency cut).
- ×Claiming you invented something that was actually a standard, well-known pattern.
- ×Failing to show external awareness - reinventing a wheel that another team or open-source project already solved.
Follow-up strategies
Interviewers will probe. Be ready for the follow-up questions that test the depth of your story.
- →If asked 'How did you know the simpler design would work?' - describe how you de-risked it (shadow mode, gradual cutover, a proof of concept against real traffic).
- →If asked 'Where did the idea come from?' - name the external source (another team, a paper, open source) to show you were externally aware rather than reinventing.
- →If asked 'What did you choose NOT to build?' - have a concrete example of scope you cut to keep the solution simple; this directly demonstrates the principle.
- →If asked 'How do you measure simplicity?' - have concrete proxies ready (lines of code, number of services or steps, on-call pages, time-to-complete a task).
- →If pushed on the tradeoff, acknowledge what your simpler version gave up and why that was the right call for the actual requirements.
Related behavioral themes
Tech Debt Pragmatism
GeneralThe judgment test. Can you ship fast when speed matters, say no to over-engineering, AND make the case for paying down debt when it's earning interest - or do you have a single mode?
Dive Deep
Amazon LPLeaders operate at all levels. The interviewer is testing whether you actually understand your own systems - or whether you summarize what your team built.
Companies that test this theme
Practice these stories live
Reading STAR answers is the floor. The interview signal is in delivering them out loud, with follow-ups, under pressure. The AI mock interview probes your stories the way real interviewers do.
Start an AI mock interview →