Hi HN. I built this during my own job search and wanted to share it.
After being laid off earlier this year, I spent a lot of time preparing for interviews. I kept running into two issues: I could not reliably tell whether my resume matched what a role was asking for, and I struggled to structure behavioral answers clearly under pressure. I talked to several recruiters and HR professionals to understand what signals they actually look for. Their feedback ended up shaping the scoring logic, question themes, and the prompting constraints for the tool.
The result is Primly.io. You upload your resume and the job description, and it gives you a match score based on criteria recruiters highlighted, like missing responsibilities, unclear ownership, mismatched seniority language, or lack of impact. It also generates behavioral interview questions and sample answers based on your resume and the role. The “tell me about yourself” generator follows a structure recruiters told me works well for early interview stages.
A few technical notes:
• The Match Score is rule based. I experimented with ML but found that simple text extraction, clustering, and weighted heuristics aligned better with how recruiters described their screening process.
• Behavioral answers use a small LLM with strict prompting to stay close to the user’s actual experience. I tried to reduce generic output as much as possible.
• The front end is built with Next, and the backend is serverless. Nothing is stored or logged. Everything happens in memory.
• I am considering moving the Match Score to run entirely client side if there is interest.
The core features are free. I added a paid option later for people who want to track a larger number of job applications, but the main functionality does not require it.
If you try it, I would appreciate feedback on anything, especially:
• whether the Match Score feels meaningful
• whether the behavioral questions reflect what you have seen in actual interviews
• any concerns around privacy or implementation
• ideas for improving the parsing or scoring
• any missing features that would make this more useful
Thanks for taking a look. I built this mostly to help myself feel less overwhelmed during interviews, but if others find it useful, that would mean a lot.
radu_trandafir•2mo ago
After being laid off earlier this year, I spent a lot of time preparing for interviews. I kept running into two issues: I could not reliably tell whether my resume matched what a role was asking for, and I struggled to structure behavioral answers clearly under pressure. I talked to several recruiters and HR professionals to understand what signals they actually look for. Their feedback ended up shaping the scoring logic, question themes, and the prompting constraints for the tool.
The result is Primly.io. You upload your resume and the job description, and it gives you a match score based on criteria recruiters highlighted, like missing responsibilities, unclear ownership, mismatched seniority language, or lack of impact. It also generates behavioral interview questions and sample answers based on your resume and the role. The “tell me about yourself” generator follows a structure recruiters told me works well for early interview stages.
A few technical notes: • The Match Score is rule based. I experimented with ML but found that simple text extraction, clustering, and weighted heuristics aligned better with how recruiters described their screening process. • Behavioral answers use a small LLM with strict prompting to stay close to the user’s actual experience. I tried to reduce generic output as much as possible. • The front end is built with Next, and the backend is serverless. Nothing is stored or logged. Everything happens in memory. • I am considering moving the Match Score to run entirely client side if there is interest.
The core features are free. I added a paid option later for people who want to track a larger number of job applications, but the main functionality does not require it.
If you try it, I would appreciate feedback on anything, especially: • whether the Match Score feels meaningful • whether the behavioral questions reflect what you have seen in actual interviews • any concerns around privacy or implementation • ideas for improving the parsing or scoring • any missing features that would make this more useful
Link: https://primly.io
Thanks for taking a look. I built this mostly to help myself feel less overwhelmed during interviews, but if others find it useful, that would mean a lot.