The short version
Taylor.io started from a simple frustration: applying to jobs felt messy, repetitive, and weirdly hard to manage.
I wanted to build a tool that could take someone's existing resume, parse it into clean structured data, and help them tailor it for specific roles without starting from scratch every time.
The project became a full-stack resume workflow with parsing, editing, previewing, and export logic built around a guided user experience.
The problem
Most resume tools either feel too rigid or too generic. They let you fill in sections, but they do not really help you understand how your experience should shift depending on the role.
The harder part was not just generating text. It was turning messy resume files into structured profile data that could be edited, reused, and exported cleanly.
What I built
- resume upload and parsing
- structured profile editing
- job-description-aware tailoring
- live preview flow
- PDF/DOCX export logic
- authentication and saved user data
Design decisions
The biggest design choice was to make the app feel guided instead of blank.
I did not want users staring at an empty editor wondering what to change. The interface needed to show them what was already there, what could be improved, and how their resume changed based on the job they were targeting.
That led to a layout focused on structured sections, live preview, and small editing decisions instead of one giant text box.
Technical decisions
The backend was built around structured data instead of treating the resume as one big document. That made the system easier to edit, preview, and eventually export.
I used FastAPI for the API layer, PostgreSQL for storing profile data, and React for the editing experience. The AI layer was used for extraction and drafting, but the important part was keeping the output structured enough that users could actually control it.
Challenges
The hardest part was balancing automation with control.
If the AI changed too much, the user could lose trust. If it changed too little, the tool was not useful. So the product had to guide the user without making the whole process feel like a black box.
The outcome
Taylor.io became a working product that helped me explore AI-assisted document workflows, structured editing, and export systems. More importantly, it gave me a clear example of how I think about software: not just as features, but as a guided system that helps people make better decisions.
What I'd improve next
The next version would focus more on onboarding, clearer version history, and a stronger way to compare tailored resumes side by side.
I'd also like to improve how the system explains its suggestions, so users understand why something changed instead of just accepting generated text.