01 / The problem
Challenge
Enterprise AI systems need clear boundaries, repeatable workflows, and human oversight—not just a model call.
02 / System view
Architecture at a glance
A simplified, public-safe model. Select a layer to inspect its responsibility.
Selected layer
Source code
Large COBOL codebases and supporting documentation.
03 / Engineering decisions
Approach
- Design Node.js and NestJS backend services for code parsing, refactoring orchestration, migration workflows, and integrations.
- Engineer LLM pipelines for source code and documentation beyond individual model context windows.
- Integrate OpenAI models into repeatable workflows with validation, retries, failure handling, and batch processing.
04 / Tools & skills
What I used
Tools
Skills
05 / Evidence
Outcomes
- Resume-reported approximately 40% less manual engineering effort.
- Resume-reported approximately 30% fewer defects.
Client engagements and proprietary platform details are intentionally kept at a high level.
06 / Reflection
What I learned
01
Legacy modernization is a systems problem as much as a model problem.
02
Validation and failure handling make AI workflows repeatable.