Wedinio Patdu

Built large‑scale architectural datasets for AI‑driven image, massing, and 3D model generation. Curated and labeled architectural images, drawings, and 3D assets with metadata. Developed AI pipelines for image and 3D generation using PyTorch, TensorFlow, and Hugging Face. Trained diffusion models, GANs, and transformer‑based models for conceptual design. Integrated text‑to‑image, image‑to‑image, and image‑to‑3D workflows. Developed a 3D environment comparable to Rhino using Three.js + OpenStreetMap.

– AI 3D House Design Platform (remote)

Built data pipelines converting IFC/DWG ↔ Revit families, enabling cross‑platform BIM collaboration. Wrote clean, scalable C# code aligned with Revit API best practices; maintained plugin documentation. Integrated AI‑based 3D model generation with architectural design systems. Built interactive web‑based 2D floor plan and 3D modeling editor using Three.js. Configured databases of furniture and materials for integration with BIM and AI systems. Developed next‑generation C# Revit plugins for AI‑driven house design automation. Created modules: IFC2USD, JSON2IFC, JSON2DWG, IFC2DWG, DWG2IFC; tested models using Solibri.

Urban building design using geospatial tools (Trian3Dbuilder, Unreal Engine) with roads, buildings, trees, railways from DEMs and satellite imagery.

Engineered scalable trading infrastructure supporting HFT, LFT, and autonomous agent‑based strategies – increased reliability by 40% under load. Designed predictive models using time‑series forecasting and statistical arbitrage; automated trade execution with 95% accuracy and 30% latency reduction. Built AI agents with CrewAI and LangChain for task coordination, tool invocation, and autonomous decision‑making. Managed high‑throughput data pipelines processing live multi‑asset market feeds (sub‑100ms latency). Containerized AI microservices using FastAPI and Docker, deployed on Kubernetes clusters on AWS/GCP with auto‑scaling and GPU support. Integrated LLMs via LangChain for autonomous trade analysis and execution through eTrade APIs. Applied CI/CD pipelines and validation workflows, increasing deployment confidence and model reliability by 80%.