EXL transforms code migration for leading UK financial institution
The client, a prominent UK financial institution, operates in a highly regulated industry with critical dependencies on robust risk and data management systems. The institution faced the pressing challenge of migrating its extensive codebase, a foundational element in its risk management and data processing workflows, to Python. The scope involved migrating various ETL jobs, scoring pipelines, and complex statistical models distributed across multiple teams.
The estimated timeline for a manual-only migration was daunting, and carried significant resource costs. The client required a scalable, secure, and efficient solution to meet their business-critical needs while maintaining compliance with internal standards and industry regulations.
Challenge
The client's challenges revolved around three critical pain points during their code migration initiative:
- Manual complexity and inefficiency: The sheer size and complexity of the risk system's SAS codebase, combined with limited documentation, posed major difficulties for manual migration, requiring considerable manual effort and advanced expertise.
- Lack of standardization: Ensuring the converted Python code adhered to the institution’s stringent coding standards was non-negotiable but impossible to guarantee in a purely manual approach.
- Time and cost constraints: A manual migration would not only demand a lengthy timeline but also extensive investments in skilled resources proficient in both SAS and Python.
These factors underscored the necessity of an intelligent, AI-driven solution that could scale efficiently and maintain accuracy while reducing time-to-completion.
Solution
EXL deployed its Code Harbor™ platform, a generative AI-powered solution designed to streamline code migration while enhancing governance and optimization. The technology was tailored to the client's specific environment through a phased implementation approach:
Pilot deployment
To validate the feasibility and efficiency of the solution, EXL piloted Code Harbor™ within EXL’s secure environment. Sample SAS codes were successfully converted to Python, demonstrating the platform's capability to meet the client's requirements for accuracy and compliance.
Full configuration with the client's environment
Post-pilot, the Code Harbor™ solution was configured to align with the client’s infrastructure, incorporating secure container deployment and tight integration with their approved machine learning models.
Comprehensive solution features
The following key features enabled seamless code migration for the client:
- Multi-agent architecture for modular code assessment, transformation, testing, and optimization.
- End-to-end code migration capabilities, including the ability to process extensive code bases without losing context through intelligent chunking.
- Compliance-driven transformations, ensuring all Python outputs adhered to the client’s coding standards.
- Versatility, with integration capabilities across on-premise, cloud, and hybrid environments.
Collaborative support
EXL’s data and AI experts worked closely with client teams to facilitate knowledge transfer, implement best deployment practices, and enable ongoing, autonomous usage of Code Harbor™ to scale migration efforts.
Results
Through EXL's cutting-edge approach and the deployment of Code Harbor™, the client achieved exceptional results:
- 70% reduction in code conversion time by leveraging advanced AI for efficient context understanding.
- 40% efficiency gain in transitioning from manual efforts to AI-assisted migration.
- Seamless adherence to the bank's internal coding standards and practices, mitigating risks of non-compliance and ensuring high-quality outcomes.
The client successfully executed the migration of modularized and scalable Python code, optimized for performance, rather than a one-to-one monolithic SAS-to-Python translation. These outcomes supported the institution's critical need to modernize while maintaining operational reliability.