An intelligent job-shop scheduling engine combining classical heuristics, evolutionary optimization, and autonomous supervisor agents for real-world manufacturing systems.
FlowForge AI delivers dynamic, near-optimal scheduling using:
- ✔ Shortest Processing Time (SPT)
- ✔ Earliest Due Date (EDD)
- ✔ Genetic Algorithm optimization
- ✔ Real-time machine failure handling
- ✔ Gantt chart visualization
- ✔ Optional FastAPI backend
- ✔ Kaggle notebook version
Classical priority rules:
- SPT (Shortest Processing Time)
- EDD (Earliest Due Date)
Useful for quick baseline scheduling.
Evolutionary optimization featuring:
- Crossover
- Mutation
- Tournament selection
Produces significantly lower makespan than heuristics alone.
Simulates real manufacturing disruptions:
- Detects OK / FAIL machine states
- Initiates auto rescheduling
- Saves state snapshots in history
Generates visual timelines of:
- Machine usage
- Job ordering
- Processing intervals
/upload_jobs
/schedule
/ga_schedule
/supervisor_step
/history
shop-floor-agent/
│
├── agents/
│ ├── scheduler.py
│ ├── ga_optimizer.py
│ ├── supervisor.py
│ └── machine_sim.py
│
├── tools/
│ ├── csv_tool.py
│ └── gantt.py
│
├── data/
├── memory/
├── scripts/
├── deployment/
│ └── app.py
└── notebooks/
pip install -r requirements.txtuvicorn deployment.app:app --reloadhttp://127.0.0.1:8000/docs
Includes:
- Job dataset loader
- SPT + EDD scheduling
- GA optimization
- Supervisor simulation
- Gantt chart rendering
Runs fully offline.
- Multi-objective GA (makespan + tardiness + energy)
- Machine setup/changeover times
- Worker skill matrices
- OR-Tools constraint solver
- Streamlit dashboard
FlowForge AI – Shop Floor Intelligence System
© 2025 Shehzan Khan. Created as a personal portfolio project.