Accelerating data modernization and tech debt reduction with AI-enabled delivery

This privately held investment management firm specializes in a comprehensive suite of investment products, risk management solutions, and advisory services. Managing funds for investors and institutions worldwide, the company employs associates globally.

The organization faced significant technical debt across legacy data platforms, compounded by large-scale migration initiatives spanning cloud providers and analytics tools. Much of the existing logic was embedded in complex SQL procedures and legacy systems, with documentation that had evolved over time, making migrations slower, higher-risk, and more dependent on a small number of subject-matter experts.

In addition, strict client constraints around tooling (such as mandated use of data build tools and Spark) limited architectural flexibility. Manual coding, documentation, and validation processes created delivery bottlenecks, slowed system upgrades, and made knowledge transfer difficult as teams inherited work from other groups.

The team incorporated GenAI and agentic workflows directly into their development and migration processes. AI was used not only to generate code, but also to reason over existing logic, automate documentation, and support iterative validation. By embedding AI tools alongside familiar developer environments and CI/CD workflows, the team accelerated delivery without disrupting existing standards or client constraints.

GenAI was applied to systematically reduce technical debt by automating code migration between languages and platforms, while simultaneously generating high-quality documentation in Confluence. In one example, complex SQL procedures were simplified from over 1,400 lines of code into concise, easily understandable outputs, dramatically improving readability and maintainability.

To further scale delivery, the team built a custom Python-based command-line tool that leverages the GitHub Copilot API to generate standardized templates (HTML, markdown, YAML, Python) and source-to-target mappings. These artifacts are automatically published to Confluence, ensuring documentation stays current with the codebase. Agentic workflows orchestrated across PyCharm, Python, Jinja templating, and Copilot enabled efficient data migrations to AWS and Databricks while optimizing cost, performance, and developer productivity.