AIG Insurance: EDSAT Modernisation

Cloud-Native, AI-Enabled Modernisation on AWS and Snowflake

As Lead Big Data Engineer, led a cross-functional team modernising AIG's legacy EDSAT platform into a cloud-native, AI-enabled architecture on AWS and Snowflake, supporting insurance claims and loss analysis.

Reverse-engineered legacy Sybase and PostgreSQL databases, SQL stored procedures and ETL logic to define the target architecture and migration strategy. Designed conceptual, logical and physical data models (star and snowflake schemas) with standardised audit columns, including timestamps, source system, batch ID, record hash and active flags, for governance, change tracking and auditing.

Designed the AI architecture using Claude Opus 4.8, LLMs and model training to automate source-to-target mapping (STTM) generation, validated with EXL's proprietary Raptor tool, cutting manual mapping effort.

Built batch and real-time pipelines with AWS Glue, EMR, EC2, S3, Python, SQL, Spark, Kafka, NiFi, Flink and Airflow on AIG's proprietary Atlantic Framework. Engineered Snowflake with Snowpipe, Streams & Tasks, Snowpark and RBAC, integrated NoSQL sources, and used Snowflake Cortex for GenAI and RAG-driven insights.

Implemented governance and lineage with AWS DataZone, AWS alerts and monitoring for SLA tracking, and ServiceNow for incident and change management. Automated CI/CD with Jenkins (CloudBees) and GitHub, and built data quality and reconciliation controls to accelerate migration sign-off.

Led and mentored engineers, analysts and QA, assigning and tracking tasks on Rally boards in Agile sprints, and partnered with stakeholders to deliver on schedule.

• Snowflake Cortex
• GenAI & RAG
• LLMs & Model Training
• Claude Opus 4.8
• Snowflake (Snowpipe, Streams & Tasks)
• Snowpark
• AWS Glue
• AWS EMR, EC2, S3
• Apache Spark
• Apache Kafka
• Apache Flink
• Apache NiFi
• Apache Airflow
• AWS DataZone
• Jenkins (CloudBees)
• Python & SQL