Ticketly: AutoML-Driven Automation
A high-performance operational AI layer that automates ticket classification and escalation using AutoML. Reduces manual triage by up to 60%.
Impact
30-60%
Impact
0.88
F1 Score
Impact
Reduction in manual ticket triage effort across support teams. Production-grade F1 score achieved with less than 1,000 labeled samples.
Process
The architecture focuses on turning unstructured text (title + description) into actionable decisions through a multi-layer pipeline.
- Preprocessing: Cleaning, feature consolidation, and PII stripping for privacy.
- AutoML Engine: Automated model selection and hyperparameter tuning for speed-to-value.
- Action Loop: Real-time routing and SLA handling with human-in-the-loop feedback.
Methods
- AutoML
- NLP Embeddings
- Text Classification
- F1-Score Optimization
- Real-time Routing
