AI Engineer project
Build a resume feedback tool with reliable structured output
A tool that scores a resume against a job description and returns structured, specific feedback — the hard part is the output being reliably parseable, not the prompt idea.
The brief
Build a tool that takes resume text and a target job description and returns structured feedback: missing keywords, specific weak bullet points each paired with a suggested rewrite, and an overall score with reasoning. The output must be reliably structured (valid JSON matching a defined schema) so it could be consumed by another system, not just readable prose.
Suggested stack
What you hand in
- A public GitHub repository with the running tool and a README
- The README documents the output schema and how malformed model output is handled
- At least two example runs (input resume/JD and the resulting structured output) included in the repo
Grading happens against the rubric below, so read it before you start — not after.
How this is graded
Published in advance and weighted out of 100. Nothing here is a surprise.
The tool validates the model's output against the schema and retries or repairs on a malformed response, rather than crashing or passing bad JSON downstream.
Feedback names the specific bullet and the specific problem with it, with a concrete rewrite suggestion, not generic advice that could apply to any resume.
The prompt reliably produces the intended structure and tone across multiple different resumes, not just the one example it was tuned against.
An empty or very short resume and a very long resume both produce sensible output rather than an error or truncated, wrong analysis.
No console errors or crashes, no broken layout, no leftover placeholder text or commented-out code.
Why this project is worth your weekend
- This maps directly onto ReachRole's own product, so the skill built here is immediately transferable to real resume-tooling work.
- Structured-output reliability is the actual hard part of shipping any LLM feature into a real system — the prompt idea itself is rarely the bottleneck.
- A tool that sometimes returns broken JSON is a tool that breaks whatever real integration depends on it, which is exactly what this project is graded on.
Where people lose points
- Parsing the model's JSON output with no validation or retry, so a single malformed response crashes the whole app.
- Feedback so generic it could apply to any resume ("add more detail", "use stronger verbs") with no reference to the actual bullet text.
- No handling for a resume far longer than the model's practical context, leading to silently truncated or wrong analysis.
Other AI Engineer projects
Two or three of these turn an empty resume into a portfolio.
Built it? Get it scored against this rubric.
Submit your work and get a score on every criterion above, written feedback, and three resume bullets you can use straight away.