Open to opportunities

andy l

@andyl

Message

ML engineer specializing in AI and machine learning optimization.

South Korea

Message

What I'm looking for

I am looking for opportunities that foster innovation and collaboration in AI development.

I am a dedicated ML engineer with a strong background in developing and optimizing machine learning systems. Currently, I work at Maum AI, where I have established a robust LLM server stress test and developed SLLM inference models that leverage advanced GPU capabilities. My experience includes collaborating with Qualcomm under a mentorship program, which has enriched my understanding of cutting-edge technologies in AI.

Previously, I contributed to a company-wide MLOps system that automated model training and deployment, achieving significant improvements in efficiency. My work on a multimodal model for a government project showcased my ability to integrate heterogeneous computing solutions. I am passionate about sharing knowledge, having conducted weekly lectures on AI models to various teams, fostering a collaborative environment.

Experience

Work history, roles, and key accomplishments

Current

MLsys engineer

Current

Maum AI

Jan 2024 - Present (1 year 5 months)

Established LLM server stress testing based on Poisson distribution to determine maximum throughput across various GPUs. Developed SLLM inference for models like Qwen, Llama 2, and Llama 3 on NPU with 4GB memory through optimization. Designed and implemented an LLM training framework to achieve faster training times compared to Megatron-LM.

LLM PyTorch Accelerate C#)Python

ML engineer

Company

Oct 2022 - Dec 2023 (1 year 2 months)

Trained models for an illegal product inspection system, achieving 87% accuracy in real-world scenarios. Built a Level 2 MLOps system utilizing Kubeflow, Kubernetes, Jenkins, MLflow, and Grafana for ML job automation and CI/CD. Designed a multimodal model for heterogeneous computing on NPU and PIM under a government project, developed with ONNX-MLIR and ONNX Runtime.

MLOps Kubeflow Kubernetes Jenkins MLFlow Grafana C#)Python Model Training

ML engineer

Triplet

Mar 2022 - Jul 2022 (4 months)

Trained in PyTorch programming and streaming input processing for AI services. Provided weekly lectures on AI models to marketing and planning/implementation teams. Built a fashion cloth object tracking and recommendation system using AWS EC2, Flask, PyTorch, and ONNX.

PyTorch AWS EC2 Flask Onnx Python Recommendation Systems

Education

Degrees, certifications, and relevant coursework

Hanyang University

Bachelor, Computer Science

Studied Computer Science at Hanyang University, focusing on core principles and advanced topics within the field. Gained foundational knowledge and practical skills relevant to software development and machine learning.