Enterprise-Grade Intelligence Engine designed for leading Ethiopian coffee exporter
# Naqel Inspection Intelligence System
---
## Description
This project is an Enterprise-Grade Intelligence Engine designed for leading Ethiopian coffee exporter.
It analyzes 2,300 high-fidelity inspection records to bridge the gap between warehouse quality control and financial profitability.
The system automates the grading of coffee lots based on:
- ECTA (Ethiopian Coffee and Tea Authority) standards
- Naqel (CLU) export inspection standards
This provides real-time visibility into export readiness and operational performance.
---
## The Problem: The $1.05M Inventory Block
Romina PLC faced a major operational bottleneck:
- 17.6% of total inventory (404 lots) failed Naqel export inspection
- $1,050,000.00 in capital was trapped in rejected inventory
### Key Challenges
- **Lack of Visibility**
Inability to distinguish:
- Total Losses (physical defects)
- Near-Misses (moisture issues)
- **Operational Inefficiency**
Poor resource allocation due to unclear failure drivers
- **High Volatility**
Quality instability at washing stations (Astatke Menafesha) disrupted the:
- 45-container weekly export cadence
---
## The Solution: Predictive Compliance & Recovery Modeling
A Three-Phase Intelligence System was developed to transform raw inspection data into actionable strategy:
### 1. Compliance Engine
- Automated grading logic for:
- Grade 1 (Specialty)
- Grade 2 (Premium)
- Instantly flags non-compliant lots
### 2. Forensic Quality Audit
- Conducted multivariate correlation analysis
- Key insight:
- Primary defects (-0.535 correlation) are the main driver of rejection
- NOT farm altitude
### 3. The $215K Recovery Model
- Built a "Rescue Filter"
- Identified:
- 83 high-quality lots failing only moisture criteria
- Recovery opportunity:
- $215,717.83 in blocked capital
---
## Strategic Recommendations
Based on Deep Audit:
### Targeted Re-Drying
- Re-process the 83 "Near-Miss" lots
- Achieve < 11.5% moisture threshold
- Unlock $215K revenue
### Station Stab …