• Design and deploy a locally hosted LLM-powered agent for autonomous payment failure analysis • Build internal LLM infrastructure with no external API dependency for core workflows • Develop structured pipelines for root cause identification across transaction failures • Automate Level 1 incident investigations • Generate standardized root cause analysis (RCA) reports • Optimize model performance to reduce Mean Time to Resolution (MTTR) • Develop scalable training and inference pipelines • Create reusable model components adopted across multiple use cases • Reduce time-to-deploy new AI applications • Decrease reliance on external AI APIs through internal model development • Implement monitoring systems for latency, drift, and model performance • Deploy at least two additional AI use cases (e.g., chatbot, automated reporting, issue clustering) • Ensure ≥99.9% production uptime • Maintain inference latency within defined thresholds • Establish retraining cadence and continuous performance evaluation • Deliver measurable efficiency improvements in operational workflows • Implement version-controlled datasets and model versioning • Define evaluation benchmarks (precision, recall, accuracy thresholds) • Implement automated drift detection • Document model architecture and training processes • Ensure zero preventable production-critical failures due to model design • Ensure personal information of customers, employees, and other individuals processed and protected in line with applicable data privacy policies.