ROLAND LI
Hi, I'm Roland. I'm a machine learning and software engineer with a solid background in deep learning and backend development.
About
Hi, I’m Roland. I’m a machine learning and software engineer with a solid background in deep learning and backend development. I’ve had the chance to work at several early-stage tech startups, where I learned how to solve complex problems quickly and effectively. Collaboration is really important to me, both in growing my technical skills and building meaningful professional relationships.
Currently, I’m prototyping a top-down survivors-style game in Unity and even integrating large language models to enhance dialogue dynamically. I love combining creativity with technology to build innovative solutions that help users, like VFX artists, save time and have greater control over their workflows.
- 2 years' experience
- Joined July 2025
Industry experience
- Software & Internet
- Gaming
- Media & Entertainment
- Retail
Work experience
Integrated the latest segmentation research in PyTorch into a quick rotoscoping product, significantly saving VFX artists time. Incorporated particle-based point tracking, object detection, and embedded text prompts with CLIP to enhance video segmentation capabilities. Designed and developed a backend microservices architecture using Python/FastAPI, relational databases, and job scheduling for a quick AI rotoscoping web app and AI plugin suite for After Effects. Managed product and ML model deployment, built load balancer, created task dispatcher, and set up custom EC2 autoscaling on AWS. Regularly documented workflows, potential bugs, and results.
Led development of a hole-detection algorithm for retail clothing images in a virtual try-on pipeline. Trained GAN-based inpainting model from scratch in PyTorch, modified Latent Diffusion architecture for inpainting, finetuned VAE and diffusion models on large retail dataset. Trained CNN-based hair segmentation model on augmented hair datasets. Containerized and deployed ML models using Docker and Kubernetes on AWS and GCP. Automated evaluation and deployment with GitHub Actions as part of CI/CD. Tracked model training with Weights and Biases and monitored performance with Grafana. Frequently communicated results and issues to cross-functional teams.
Education
Hire ROLAND LI today
Post a job and invite ROLAND, or contact them directly. No hidden commission.










