Sude Naz Karabacak completed AI-Assisted Code Modernization with IBM

Built and verified a full legacy-code modernization for a vehicle maintenance service: fixed wear calculation and missing-data bugs, repaired nightly reporting, added regression tests, corrected a hidden unit-conversion defect in helper code, and produced a data-driven breakdown-risk analysis. The work preserved the 15,000 km / 80% business rules, passed the acceptance checks, and documented the AI review process clearly.

• Leakage-Free Feature Engineering
• Honest Model Evaluation
• AI Agent Prompt Scoping
• Verification-Driven Debugging