The call, version 2 — what changes
The shape of the call, the project story, the behavioural skeletons and the phrase bank are the same as version 1’s call script. This file has only what changes for this ad: the self-intro, the questions it adds, and your questions for them.
Your self-intro — 90 seconds
Same resume, same two projects, different emphasis. This ad scores engineering practice, so the intro names the practices inside the first minute.
“Thanks. So — I’m a Python backend engineer with over five years of commercial experience, building API services and integrations, mostly on AWS.
Most recently I worked on an automation platform for underwriting — serverless Python on AWS: Lambda services orchestrated by Step Functions, state in DynamoDB behind API Gateway, AI extraction, and REST integrations with Salesforce and OAuth2 services like Microsoft Graph and Slack. I also worked on the delivery side — GitHub Actions workflows building SAM applications and pushing images to ECR.
Before that, an investment planning platform — a regulated financial product: FastAPI services on PostgreSQL, data models including audit and consent records, third-party integrations, and test-driven development with pytest, unit and integration.
The way I like to work: typed Python checked in CI, small pull requests on short-lived branches, tests written first, and changes I can observe once they’re in production. My last two products were regulated finance and underwriting, where careful data handling is part of the job — so a healthcare product feels like a natural next step. Happy to go deeper on any of that.“
Why it is built this way. The last paragraph names four of the ad’s requirement lines — typing, Git workflow, testing, observability — in one breath, so the interviewer’s checklist starts filling before the first question. Everything else matches the resume they are holding, in its own words.
The project story
Tell the same two projects the same way — version 1’s project story has the shape, the two-project table and the drill-downs. For this ad, lean on these threads:
- Automation Platform — the CI/CD and infrastructure thread (GitHub Actions, SAM, ECR), how the Step Functions workflow handled a failed step, and how the Salesforce and OAuth2 integrations were tested.
- Invest Planning Platform — the testing thread (TDD with pytest, unit and integration), the FastAPI API design, and the audit and consent records for the regulated question.
The questions this ad adds
“How do you use AI tools?”
The ad names it, so it will come up. The spoken answer is in Answers, version 2 — say these out loud. What makes it land is being specific:
“[The tools you use, by name, and what each is for.] [One you stopped using, and why.] [A time AI-generated code looked right and wasn’t, and how you caught it.]”
“Tell me about a performance problem you solved”
Symptom → measurement → cause → fix → the number after.
“[What was slow, and for whom.] I measured it with [the trace, the profiler, the query plan], and the cause was [the cause]. I [the fix], and [the metric] went from [before] to [after]. Afterwards we [added an alert, a load test, a regression check] so it would not come back quietly.”
“Tell me about a change that broke production”
The ad’s CI/CD line, asked as a story. Version 1’s incident skeleton fits; end on the gate you added to the pipeline afterwards — the test, the check, the canary step — because that is the part this ad scores.
“Have you used Kafka / Kubernetes / Terraform?”
The preferred list has several things your resume does not. The answer that works is honest, then mechanical, then bridged:
“Not in production. [One sentence on how it works — the partition key decides ordering; liveness touches nothing, readiness checks dependencies; plan in CI, apply from the pipeline.] The closest thing I’ve run is [SQS and Step Functions / ECS / SAM], where [the same idea showed up as…].”
Then stop. Over-explaining a gap sounds less confident than naming it.
Questions to ask them
Ask two or three:
- “Which of the preferred list is real day to day — Kafka or SQS, Kubernetes or ECS, Terraform or Pulumi?”
- “Is any of the software regulated as a medical device — and if so, how does that change the release process?”
- “How do services test against each other today — consumer-driven contracts, schema checks, or shared environments?”
- “What does the path from merge to production look like, and how long does it take?”
- “How does the team use AI tools — and are there rules about what may go into them?”
The close and the six phrases are the same as in version 1.