FULL-STACK / LLM-DRIVEN ANALYSIS / 2026
Job Pilot
Built for the messy middle of job searching.
Job Pilot uses full-stack workflows and LLM analysis to help organize applications, understand role requirements, surface skill gaps, and plan what to do next. Built end-to-end (46 API endpoints, tested backend, working frontend) and used to run my own job search.
- API Design
- Prompt Engineering
- Structured Extraction
- User Flows
THE BUILD, IN NUMBERS
view code on github →46
API endpoints
FastAPI routes across the job-search workflows.
5
data models
SQLAlchemy models with Alembic migrations.
44
backend tests
pytest coverage over parsing, matching, and planning.
8
app screens
Next.js frontend pages, board to interview prep.
TECH STACK
FRONTEND
- Next.js
- React
- TypeScript
BACKEND
- FastAPI
- Python
- Pydantic
DATA LAYER
- SQLAlchemy
- SQLite
LLM LAYER
- Prompt design
- Structured outputs
- Role analysis
- Gap analysis
THE PROBLEM
"What should I do next in my job search?"
Instead of scattering postings, resumes, notes, and interview prep across a dozen files, Job Pilot keeps the whole workflow in one place - and always surfaces the next move.
CORE WORKFLOWS
01 / 0501
Track every application
Roles, companies, statuses, notes, and next steps stay on one board.
02
Parse messy job posts
Paste a description and extract responsibilities, qualifications, and required skills.
03 / 0503
Match resume to role
See what lines up, what is missing, and what to sharpen.
04
Turn gaps into a plan
Weak or missing skills become a focused, ordered study plan.
05
Prep for the interview
Role-specific questions and review topics, pulled from the saved job.
WHAT I FOCUSED ON
- API routes designed around real job-search workflows
- Data models for applications, resumes, posts, and parsed role details
- Pydantic schemas for structured, predictable requests and responses
- AI features that produce reviewable, editable outputs
- A dashboard that turns scattered tasks into clear next steps