Client problem: Legacy PLC/HMI code that was slow and risky to troubleshoot or modify.

Solution: Built a RAG/LLM-assisted documentation pipeline that parses PLC ladder logic + HMI context into structured data and generates traceable rung/routine explanations.

Result: Legacy code AI commenting resulted in faster troubleshooting and less engineering effort for ongoing changes.

Client problem: High-uncertainty R&D project to determine if generative AI could replace manual VFX for expanding standard video into panoramic formats.

Solution: Built a working prototype, prototype genAI video results, and the strategic guidance needed to turn a speculative concept into a cost-effective production reality.

Result: Validated AI as a high-margin, low-overhead alternative to traditional VFX, and how to do it.

Client problem: Determine whether Computer Vision or AI could automate tuft-test video interpretation and produce repeatable aerodynamic insights without slow manual review under messy real-world conditions.

Solution: Delivered a prototype combining baseline CV + an ML segmentation model, additionally built a new custom scalable synthetic-data pipeline to minimize labeling and accelerate iteration.

Result: Validated tuft video → segmentation → per-tuft analytics and delivered a roadmap plus a compact segmentation model that works on real vehicle footage.