Describing your strongest project
The most predictable question in any interview, and the one most often answered as a tour of the stack. The interviewer is not collecting technologies — they are looking for a hard decision you owned and can defend under follow-up.
Pick the project with the hardest decision, not the biggest name.
The project
End-to-End Invest Planning Platform Sep 2025 - Feb 2026 | Fullstack Python / AI Developer
A goal-based wealth planning platform with a FastAPI backend and Vue 3 frontend. It covers client onboarding, portfolio and goal tracking, and financial health scoring, with AWS integrations and a compliant AI-powered financial chatbot for regulated advice.
What I delivered
- Backend: FastAPI services for RESTful APIs and microservices (onboarding, goals, financial health scoring). PostgreSQL and SQLAlchemy for data models (client, goal, fact-find, audit/consent).
- Integrations: Third-party APIs and data feeds - auth, messaging, CRM, JIRA, logging, monitoring. External API clients with auth, retries, and error handling.
- AI chatbot: Built and improved an AI chatbot with LangChain/LangGraph: information transformation, tool integration, consent flows, and financial guardrails for regulated advice.
- Testing: TDD with Pytest - unit and integration tests for reliability.
- Frontend: Vue onboarding, portfolio, and goal management screens.
- DevOps: Docker, CI/CD (Bitbucket), and observability (Sentry, logging).
Technologies: Python, FastAPI, PostgreSQL, Pytest, Celery, Redis, AWS (RDS, SNS, SQS, ECR, Cognito), Docker, Bitbucket CI/CD, Sentry, Ably, LangChain, LangGraph, TypeScript, Vue, Pinia, Tailwind, Storybook, Chart.js.
Turning that into a STAR answer
The list above is scope, and scope is not a story. An interviewer hearing it learns what existed, not what you decided. Before the interview, fill these in for one thread from the list - the regulated chatbot is the strongest candidate, because the constraint is unusual and the trade-offs are real.
| What to have ready | |
|---|---|
| Situation | the constraint that made it hard, in one sentence |
| Task | what you specifically owned, and its boundary |
| Action | the options considered and why you chose one |
| Result | what changed, measured if you can measure it |
Action and Result are where these answers usually fail. “I built a chatbot with guardrails” is a Task. The Action is which guardrail design you picked over which alternative, and why regulated advice ruled the other out. The Result is what it let the business do, or what it prevented.
If a number is genuinely unavailable, say what changed qualitatively and be straight that it was not measured. That is far stronger than an invented metric, which collapses on the first follow-up.
Questions to expect on it
- How did you design the AI chatbot for regulated advice (guardrails, consent flows)?
- How did you integrate third-party APIs (auth, retries, failure handling)?
- How did you structure the FastAPI services and data models?
- What was the hardest part of the project?
- How did TDD affect the development flow and reliability?
- How did you handle compliance and audit/consent records?
- How did you coordinate backend, frontend, and AI work?
Each of those is a place the story can go deeper, so have one concrete decision ready for each rather than a summary of the area.
Related
Interview angle 4
- “Describe your strongest project.” - pick one where you owned a genuinely hard decision, not the one with the biggest name. Structure it as problem, constraints, options you considered, what you chose and why, and the measured outcome.
- “What made it strong?” - answer in terms of impact and difficulty, not technology. “We cut p95 from 4s to 900ms and turned provider outages into degraded results instead of incidents” beats a list of frameworks.
- “What was your specific contribution?” - be precise about the boundary
between what you did and what the team did. Interviewers probe this, and
overclaiming collapses quickly under follow-up questions.
- Follow-up: “So what did the rest of the team do?” - a candidate who has overclaimed hesitates here. Knowing exactly where your work ended is itself a seniority signal, and it costs nothing to be accurate about.
- “What would you do differently?” - have a real answer ready. Volunteering a genuine limitation, ideally one whose consequences you later saw, is the single highest-value sentence in the story.