Fambai CV: Using Gemma 4 to Help African Job Seekers Beat ATS Systems

The ATS analyzer backend is built with FastAPI + Python, deployed

on Railway via Docker. The Flutter mobile app handles the frontend

with Firebase for auth and storage.

How I Used Gemma 4

I chose Gemma 4 E4B — the efficient 4-billion parameter model

— for a specific reason: I needed powerful language understanding

that could run fast enough to give users real-time feedback without

burning through API costs that would price out my African user base.

Gemma 4 E4B powers the ATS Keyword Analyzer feature. When a user

pastes a job description, the model:

1. Extracts the key skills, qualifications, and action verbs the
   employer is screening for


2. Compares them against the user's CV content


3. Returns a match score with specific, actionable suggestions

The model's instruction-following capability is sharp enough to

return structured JSON that maps directly into the Flutter UI —

no post-processing hacks needed.

For a solo founder building for users who can't afford to waste

a single application on a bad CV, Gemma 4 E4B was the right

balance of intelligence and efficiency. It's fast, it's accurate,

and it respects the constraints of building for emerging markets.

This is what AI should do — level the playing field.