Jay Anjankar

Designed and deployed a hands-free RAG system integrating Llama models and NVIDIA Riva for ASR and speech synthesis, achieving high accuracy in standard and noisy conditions. Led a team to fine-tune InternVL for visual-language tasks and developed custom plugins for 3ds Max & Substance Painter, reducing material assignment time by 70%. Developed orchestration scripts for DCC automation, enabling batch FBX processing and improving efficiency for 3D artists. Engineered a digital twin with NVIDIA Omniverse for traffic intersection analysis, achieving high object and violation detection accuracy. Built synthetic data pipelines for port operations digital twin, improving processing time and simulation accuracy. Directed a team to develop a plugin integrating Autodesk Revit with NVIDIA Omniverse, driving enterprise adoption.

Developed OCR and bounding box regression models using TensorFlow with high accuracy for automated recognition of manufacturing components. Improved software reliability by building Pytest-based unit tests for a large codebase and implementing critical bug fixes.