Jobot test release - Please Beat Up My App
I received my 50th (hundredth?) rejection email on Friday and noticed that the company hadn't even bothered to change to the rejection email text or the 'from' email address: everything was still the same as the ATS firm's template.
So, like every good indiehacker, I decided that I needed to build an app to solve this....
I wanted to compare my resume to the job description to:
Get a paper comparison - how would I look to a human
A mechanical comparison - how do I look to a machine
Recommendations on missing attributes and suggestions for where the wording in my resume might not exactly match the job description
Jobot V1
After a weekend of building and testing, I just released Jobot V1 for public testing. It addresses all the main issues and provides some nice alerts where there is a hard requirement missing. The recommendation for terminology adjustments to get exact matches are also working. I'd say this is about 85-90% there with respect to core functionality, but there's some refinement to do on output formatting and consistency.
👉 The Big Ask
I'd love for a few folks to take a look at the app and try it out with a resume and the kind of job you might apply for. You should get some reasonable and helpful results but any feedback is useful.
Note please don't include any PII. I clear out the logs so there are no user profiles being stored at the moment but also plan to use a PII filter to remove this from submssions.
Next steps
I plan to add a user profile so you can save your core resume in the app to avoid having to upload this each time.
Add something like Llamaguard to remove PII from submitted resumes.
I am looking at a rewrite function to revise your resume to better match the job description
These would probably both be paid tiers as they feel like power user features, and maybe someone pays for a month or two of advanced features while they job hunt.
PS - Replit FWT
Seeing as I only write python 🐍 and wanted to focus on the AI aspect which would be the most time-consuming, I turned to Replit. The agent built the interface with a blank assessment_model.py file and I then iterated on the assessment methodology, prompts and output mapping. 75% of the time was on the core code so of the four - five hours this took, only an hour was set up and configuration of the app. Thank you Replt ...🙏
