Building next-generation AI platform for investment banking, private equity, and venture capital at yAI (Antler 2025).

• Conducted research and optimized an open-source deep learning vehicle trajectory prediction model (PyTorch, Waymo Open Dataset), achieving 86.25% accuracy and 0.63m ADE for NYC traffic forecasting to enhance urban road safety.

• Developed a containerized vehicle detection platform using Gradio, Hugging Face Spaces, and Docker, achieving 92% accuracy with 25 FPS real-time inference and generating text-based descriptions to enhance NYC traffic monitoring.

• Developed an AWS Step Functions workflow for computer vision startup’s core product using Python and CloudFormation IaC, containerizing image processing tasks with Docker, resulting in 80% reduction in processing time.

• Engineered a real-time 3D visualization platform using Three.js and WebSocket, processing LiDAR sensor data streams to render dynamic factory operations with less than 50ms latency and supporting 5+ concurrent 3D model views.

• Led development of an automated CI/CD pipeline on AWS (CodePipeline, ECR, ECS), reducing deployment time from hours to minutes and achieving 99% deployment success rate through containerized microservices architecture.

• Engineered a multithreaded state machine on AWS using Python for parallel file processing and concurrent task execution at a computer vision startup, increasing system efficiency by 80% through dynamic resource allocation and auto-scaling.

• Architected a CI/CD pipeline integrating Bitbucket and AWS CodeBuild for automated testing and deployment, implementing automated rollback strategies, achieving less than 10-minute deployment cycles with 99% reliability.

• Spearheaded startup’s cloud migration with AWS Infrastructure as Code (IaC) using agile methodologies, architecting resilient systems with EC2, VPC, and auto-scaling that handled 2M+ monthly requests while reducing costs by 40%.