Resume Agent Next.js
An AI-powered portfolio generator that processes a resume into structured profile data and renders a reusable personal site.
2-stage
Agent workflow
The pipeline separates scanning from enhancement so extraction and presentation can evolve independently.
2 providers
AI flexibility
The project supports both OpenAI and Ollama-based local models, which makes experimentation and deployment more flexible.
Generated JSON
Structured output
Resume content is converted into profile-data.json so the site can render consistent sections from one source of truth.
Problem
Keeping a personal portfolio updated by hand is repetitive and often drifts away from the actual source resume over time.
Approach
Built a processing pipeline that ingests a resume PDF, runs a two-stage agent workflow, generates structured profile data, and feeds reusable site sections from that output.
Impact
The result is a portfolio system that can be refreshed from a resume source file instead of rewriting the entire site manually.
Overview
This project is about turning resume maintenance into a system instead of a manual editing task.
The idea is simple: use a resume as the source document, process it into structured data, and render a professional portfolio experience from that output. The real value is not the AI label by itself; it is the automation around a workflow people repeat constantly.
Core workflow
The app reads generated profile data at runtime and falls back to a default onboarding state when that data is not available yet.
The pipeline behind that experience is:
1. add resume.pdf 2. run the processing script 3. parse the resume into structured information 4. enhance and normalize the output 5. render the portfolio sections from generated data
Why the build is interesting
Agentic architecture
The README describes the system as a two-stage agent framework, which is a meaningful design choice. It keeps extraction and enhancement separate so the processing can be improved without rewriting the full flow.
Flexible providers
The project supports multiple AI providers, including OpenAI and Ollama. That makes it stronger as an engineering build because it is not locked to a single model path.
Production-minded fallback behavior
The default state is intentionally safe. If generated content is missing or invalid, the app renders a clear fallback experience instead of crashing or leaving the page unusable.
Quality considerations
The most important quality concerns in this kind of system are:
- validating the generated data shape
- failing safely when resume input is missing
- keeping output sections consistent even when extraction is incomplete
- making regeneration predictable for future updates
Why it belongs in the portfolio
This project shows a different kind of engineering depth than a typical landing page. It connects automation, AI workflow design, content structure, and production fallback behavior in a way that is easy for other people to understand and evaluate.
Project proof
Stack