Cognitive Echo (machine-learning-daily)

A daily-build project: an AI personality and behavior analyzer that infers traits from interactive micro-tasks, reaction times, and short text answers. Progress is committed every day, and the daily logs in this repo document each step.

What it does

Users complete a short assessment (binary choices, reaction-speed tasks, short text prompts, and confidence ratings). The backend records answers with reaction time and confidence, and ML models score five dimensions: Openness, Conscientiousness, Risk-taking, Emotional Stability, and Decision Speed.

Architecture

Flutter (frontend)  ->  FastAPI (backend)  ->  ML models (scikit-learn / XGBoost / S-BERT)  ->  insight report

Repository layout

backend/
├── app.py            # FastAPI app with SQLite storage
└── requirements.txt
PROJECT_PLAN.md       # Day 1: system design, question set, schema
DAY_2_PROGRESS.md     # Day 2: data collection engine
daily_update.sh       # Commit-and-push script used for daily updates
daily_log.txt         # Timestamped update log

API (current)

Method	Endpoint	Purpose
POST	/submit-response	Store an answer with reaction time and confidence
GET	/user-summary/{user_id}	Aggregate stats for a user
GET	/health	Health check

Running the backend

cd backend
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
uvicorn app:app --reload

The SQLite database (cognitive_echo.db) is created automatically on first run. Interactive docs are at http://127.0.0.1:8000/docs.

Roadmap

See PROJECT_PLAN.md for the full plan and DAY_2_PROGRESS.md for the latest completed milestone. Upcoming: feature extraction from responses, model training, and the Flutter client.

Documentation

Developer docs live in docs/:

• Architecture
• Setup
• Contributing