HyunJae Lee
Machine Learning Engineer
I am a full-stack machine learning engineer with extensive experience in both research and engineering. My research has consistently focused on solving real-world problems, which has resulted in multiple publications at top-tier conferences. I have established and expanded an MLOps team from scratch to six members, and I led the design and implementation of both a cost-effective, cloud-based ML platform and an automated AI model development framework. I was responsible for the full system design and the initial development, handling backend, frontend, and DevOps all at once. I am passionate about solving real-world problems by bridging the gap between ML research and engineering.
Work Experience
Lunit Inc.
Seoul, South Korea
Research Engineer / AutoML Team Leaader, July 2021 – Present
- Established and scaled the team to six members focused on automation and efficiency in ML model development.
- Architected and implemented a large-scale, cost-efficient ML platform, leveraging over 1000 GPUs for simultaneous training at a cost reduction of more than 50%. This platform supports nearly 100,000 experiments annually.
- Designed and developed an automated AI model development framework that streamlines the repetitive aspects of AI model creation, enhancing efficiency and consistency.
- Advanced research in Bayesian optimization and pruning algorithms for hyperparameter optimization, resulting in publications and direct integration into our operational framework.
Research Scientist / 3D Mammography Team Leader, March 2018 – July 2021
- Conducted research that addresses the domain gap in both internal and external datasets, significantly enhancing model performance and robustness, leading to publications in top-tier conferences and practical application.
- Developed a 3D Mammography cancer detection AI product from scratch that received FDA clearance.
- Built a complete pipeline including data preprocessing, model development and model deployment.
ConvIoT
Seoul, South Korea
Lead Developer / CEO, June 2016 – July 2017
- Developed and launched a user-friendly IoT platform enabling simple workflow configuration via a web interface.
- Integrated a wide range of internet services and IoT devices from diverse vendors, enabling seamless and automated interactions across systems, and secured early investment from Naver.
Selected Publications (800+ citations)
Bayesian Optimization Meets Self-Distillation
HyunJae Lee*, Heon Song*, Hyeonsoo Lee*, Gi-hyeon Lee, Suyeong Park, Donggeun Yoo
BOSS framework was developed to address a real-world challenge at Lunit, where only the best model is retained from Bayesian Optimization (BO) trials, leaving other potentially useful models abandoned. BOSS enhances model performance by combining BO with Self-Distillation (SD), leveraging insights from all training trials rather than just the top model. This approach not only suggests optimal hyperparameter configurations but also uses pre-trained models from previous trials for SD, delivering superior results across various tasks. Integrated into the INCL framework, this method has led to significant performance improvements across multiple products.
Reducing Domain Gap by Reducing Style Bias
HyunJae Lee*, Hyeonseob Nam*, Jongchan Park, Wonjun Yoon, Donggeun Yoo
Style-Agnostic Networks (SagNets) were introduced to tackle a real-world challenge in medical AI products: mitigating external style variability, where style differences in external use cases led to performance drops. SagNets reduce the intrinsic style bias of CNNs by disentangling style encodings from class categories, allowing the networks to focus more on content rather than style. This approach enhances the robustness of models across different domains. It led to a significant performance improvement of our product on the external site, proving its real-world impact.
SRM : A Style-based Recalibration Module for Convolutional Neural Networks
HyunJae Lee, Hyo-Eun Kim, Hyeonseob Nam
Style-based Recalibration Module (SRM) was developed to address a real-world issue in medical AI products: managing internal style variability, where style differences across various data sources degraded network performance. SRM addresses this by adaptively recalibrating feature maps based on style, enhancing the performance of CNNs in various vision tasks. By improving the representational capacity of CNNs, SRM seamlessly integrates into existing architectures with minimal overhead. This method led to a substantial performance improvement in our product by fully leveraging the potential of diverse data sources.
Featured Projects
AutoSCOPE Framework
AutoSCOPE is an automated AI product development framework designed to streamline and automate the creation of deep learning models within Lunit's oncology department. It simplifies the entire AI model lifecycle, from data management and model training to inference on whole slide images, through a user-friendly web interface. I was in charge of the initial design and implementation, coordinating with six different teams to standardize the process. I managed a team of six members to develop and continuously improve the framework. AutoSCOPE has already trained many real-world products, leading to collaborations with major pharmaceutical companies. It provides much faster and more scalable AI product development compared to manual model development by automating complex, repetitive tasks and integrating advanced AutoML algorithms.
Intelligent Cloud Framework
INtelligent CLoud (INCL) is a deep learning training platform that streamlines the process of training AI models on the cloud. I initially proposed this project, designing and implementing the first version of INCL. I managed a team of five members to develop and continuously improve the framework. INCL supports nearly 100,000 experiments annually, with all researchers at Lunit relying on it. By maximizing the utilization of spot instances, INCL achieves a cost reduction of more than 50%, saving millions of dollars each year. It automates manual processes in cloud-based model training, including experiment setup, tracking, and management, while providing advanced features like multi-node distributed learning and automated hyperparameter optimization. By utilizing INCL, Lunit has been able to efficiently scale computational resources (more than 1,000 GPUs at the same time), improve model performance, and accelerate the overall research process.
3D Mammography Cacner Detection
Lunit INSIGHT DBT is a medical AI product designed to detect suspected areas of breast cancer in tomosynthesis (3D mammography) images. Trained on large-scale DBT datasets from diverse countries and devices, INSIGHT DBT is optimized for various breast densities, ethnicities, and manufacturers, and has received FDA clearance. I led its development from scratch, building a complete pipeline encompassing data preprocessing, model development, and deployment. I also conducted research to address domain gaps in both internal and external datasets, significantly enhancing model performance and robustness. These research advancements are integrated directly into the product, leading to publications in top-tier conferences and improving real-world applications. INSIGHT DBT improves radiologists' diagnostic performance, reduces reading times, and offers strong performance for both fatty and dense breasts.
Selected Awards
1st place in Visual Domain Adaptation Challenge
ICCV 2019
Won first place in the famous VisDA-2019 Challenge as the main contributor, demonstrating excellence in Semi-Supervised Domain Adaptation and employing techniques to address real-world challenges in the medical domain.
2nd place in Embedded Deep Learning Design Challenge
ESWeek 2017
Achieved second place in the EDLDC 2017, demonstrating an energy-efficient deep learning design for object detection on embedded systems, utilizing advanced techniques such as reduced precision inference, tensor decomposition, etc.
Best Start-up Award
Naver Demo Day 2017
Won the Best Start-up Award at Naver Demo Day 2017 as Lead Developer/CEO of ConvIoT, for developing and launching an innovative IoT platform that simplifies workflow configuration and integrates various internet services and devices.
Technical Skills
ML & MLOps
Pytorch, Scikit-learn, Optuna, Vertex AI, MLflow, Kubeflow
Backend Development
Django, DRF, FastAPI, Gin, Celery, Firebase, MySQL, PostgreSQL, Redis, Kafka
DevOps & Infra.
Kubernetes, Docker, Terraform, ArgoCD, GitHub Actions, Prometheus, Grafana, Loki, GCP, AWS
Frontend Development
React, Next.js, MobX, Vercel, Material UI, Tailwind CSS, Shadcn
Languages
English (fluent), Korean (native)