JD-to-resume customiser
The assignment
Build a tool that takes a job description and customises your resume to match. Outputs a finished PDF — not a chat-bound Gem.
✳What Baskar shipped
Trajectory runs a multi-engine analysis comparing a resume against a job description, then opens an interactive workbench where each resume line gets a specific, JD-grounded rewrite suggestion with its own rationale that can be individually accepted, rejected, or edited. The tool flags gaps that can't be filled truthfully, exports both the analysis report and the tailored resume as PDFs, and supports saving multiple company/role analyses for reuse.
✳Our read on this build
Baskar shipped a deployed web app you can open. This read comes from the artifact and Baskar's own account of it — no scores, just what the build does and does not show.
What it demonstrates
- The output is a file, not a chat reply — a formatted PDF and export — which is the whole distance between a tool and a conversation.
- It reads the posting before it rewrites: gap analysis, tailoring and rewriting.
- And it is shaped like something you would run again: more than one role.
Take this awaySteal the framing: the deliverable defines the product. Deciding 'this outputs a PDF a recruiter can open' forces every design decision after it.
✳Why this quest matters
The output format is the product. A resume that only lives in a chat window can't be sent to a recruiter — forcing a real PDF is what turns a nice answer into a tool someone would actually use.
✳What separated the strongest builds
A finished file, not a chat answer
Something you could attach to an application: a formatted, downloadable PDF. The brief ruled out anything that only ever prints text into a chat window.
Matching that has actually read the JD
Responding to that specific posting's language and priorities and reordering real experience — as opposed to sprinkling its keywords into the same resume.
It survives the second job description
A pipeline you can run again on Monday, not one artful manual edit dressed up as a tool.
No numeric scores are published here. These pages are written reads of what each builder shipped — the build is the point, not a mark out of a hundred.