Giáo trình khóa học
Bài | Số buổi | Modules | Labs |
---|---|---|---|
Fundamental of Software Architecture | 1 | + Architecture characteristics: operational, structural and others + Architecture foundations: big ball of mud, unitary and multi-tier architectures + Architecture styles: layered, microkernel, event-driven and microservices | |
Model Development Following CRISP-ML(Q)
| 2 | + CRISP-DM methodology for data mining + CRISP-ML(Q) methodology for model development + CRISP-ML(Q) components and supplementary tools | + Create a time-series forecasting solution in Jupyter Notebook following the CRISP-ML(Q) methodology. + Develop an intrusion detection example using appropriate methodologies and tools. |
Model Training & Serving Design Patterns | 2 | + Model training design patterns + Model serving design patterns: caching, microbatching, serverless architecture, high-density model serving and more + Model-in-image vs Model-load + CPU, GPU and CUDA | + Deploying YOLOv8 on CPU using BentoML with adaptive batching, exception handling, and record-level monitoring techniques. + Deploying YOLOv8 on GPU with Triton, optimized with ONNX and TensorRT, and tuning configurations using performance and model analyzers. + Optimizing GPU training efficiency with NVIDIA Nsight System. |
Serverless Inference with KServe | 2 | + Introduction to serverless architecture + Differentiate between serverless and server architecture + KServe architecture + Cloud Native Buildpacks | + Deploy an API for Intrusion Detection on KServe with Autoscaling and Scale-to-Zero capabilities |
High Density Model Serving with Model Mesh Architecture | 2 | + Pros and cons of server and serverless architectures + Modelmesh (mm) architecture | + Rapidly deploy an sklearn model following the MM architecture. + Customize the MM serving runtime to handle multiple intrusion detection models within a single k8s pod. + Develop a custom Kubernetes controller (CRD) to facilitate autoscaling of the serving runtime. |
Kubeflow | 2 | + Kubeflow architecture + Kubeflow operators | + Explore Kubeflow operators + Create a quick-start house price prediction pipeline using Kubeflow. + Develop a comprehensive intrusion detection pipeline with four stages: data preparation, training, evaluation, and visualization (including numerical metrics and images). |
Feature Store | 1 | + Introduction to feature store + Feast and core concepts + Feature view design | + Set up a feature platform locally with PostgreSQL, Redis, and Kafka. + Implement a data ingestion pipeline to populate an offline store and synchronize with an online store. + Develop a freshen job using Kafka as a data source to update data in both stores. |
ML System Design & MLOps | 3 | + Distributed Systems + Fundamentals of low-level design + c4 model for software architecture + MLOps | + Architect a real-time power consumption monitoring system. + Design a candidate matching system based on graph algorithms. + Architect a Real-Time Face Recognition system. |
Bài giảng mẫu
Serverless Architecture
Lịch khai giảng
Các lớp sẽ đóng đăng ký khi đủ số lượng học viên, thường sẽ đóng trước lịch khai giảng 1-3 tháng
Đội ngũ giảng viên
Quân Đặng
Founder & CEO
Giảng viên khóa: MLOps
Founder FSDS & Capydata. Senior Data Scientist với hơn 8 năm kinh nghiệm trong ngành.
Quý Đinh
Senior Machine Learning Engineer
Giảng viên khóa: Recommender System
Senior Machine Learning Engineer tại Chope Singapore với hơn 7 năm kinh nghiệm trong ngành.
Đức Lương
Senior Machine Learning Engineer
Giảng viên khóa: Data Engineering
Senior Machine Learning Engineer tại GeoComply với hơn 7 năm kinh nghiệm trong ngành.
Tú Trần
Research Scientist
Giảng viên khóa: Natural Language Processing
Research Scientist tại Knorex với hơn 5 năm kinh nghiệm trong ngành.
Hoàng Phạm
Head of AI OLLI Technology & Lecturer ĐH KHTN TPHCM
Giảng viên khóa: Computer Vision
Có hơn 13 năm kinh nghiệm trong ngành.
Quan Huỳnh
Data Science Manager
Giảng viên khóa: Machine Learning
Đang làm việc tại HCLTech có hơn 8 năm kinh nghiệm trong ngành.
Tiến Hoàng
Lead Data Analyst
Giảng viên khóa: Data Analytics
Đang làm việc tại Property Guru có hơn 6 năm kinh nghiệm trong ngành.
Nguyên Phạm
Senior MLOps Engineer
Giảng viên khóa: MLOps
Đang làm việc tại MoMo với hơn 5 năm kinh nghiệm trong ngành.
Câu hỏi thường gặp
Trở thành chuyên gia
Machine Learning Engineering
Tham gia cùng hàng trăm học viên toàn cầu và cùng phát triển với fullstackdatascience!