Greenlight

"Get the greenlight to apply."

Technical portfolio gap advisor for data and analytics engineers.

How it works

persona.md + job offer URLs
  → fetch + skill extraction
  → DuckDB + dbt (staging → marts)
  → Python gap scorer
  → LLM recommendation (Mistral)
  → Streamlit UI output

Stack

Layer	Tool
Frontend	Streamlit
Transformation	dbt-duckdb
Local warehouse	DuckDB
Skill extraction	Python + BeautifulSoup
LLM hook	Mistral AI

Local setup

Requirements: Python 3.10+

# 1. Clone and create virtual environment
git clone https://github.com/your-username/greenlight.git
cd greenlight
python -m venv .venv
.venv\Scripts\activate        # Windows
# source .venv/bin/activate   # macOS/Linux

# 2. Install dependencies
pip install -r requirements.txt

# 3. Configure environment
cp .env.example .env
# Edit .env and add your Mistral API key:
# MISTRAL_API_KEY=your_key_here

# 4. Run the app
streamlit run app/streamlit_app.py

Usage

1. Open the Streamlit app in your browser.
2. Upload your persona file (see persona/hamza.md as reference).
3. Paste job offer URLs (one per line, or use offers.md format).
4. Select LLM mode and click Run.
5. View your top skill gaps and the recommended next project to build.

Persona format

Create a Markdown file following this structure:

## Skills
# tool | proficiency (1-3) | used in production?
dbt | 2 | yes
Python | 3 | yes

## Projects built
# name | domain | pipeline stages covered | dbt concepts used
my-project | fintech | ingestion+transform+test | surrogate keys, window functions

## Target roles
Analytics Engineer, Data Engineer

## Domain preferences
fintech, marketing analytics

## What I want to learn next
DuckDB, LLM pipelines

## Time available for a project
~2 weeks

CLI usage

python run.py "https://wttj.co/offer1" "https://wttj.co/offer2"
# or pass a file of URLs
python run.py offers.md

Project structure

app/              Streamlit frontend
ingestion/        Fetch, parse, and load offer data
scoring/          Gap scoring logic and LLM hook
transform/        dbt project (staging → marts)
persona/          Persona Markdown files
data/             DuckDB database (gitignored, created at runtime)
offers/           Cached offer text and extracted skills (gitignored)