Accelerating End-to-End Data Science Workflow

Course Code: IT 1885

4 Course Visits

Course Sector:

Information Technology

Request in house training Call Me Back

Course Dates and Locations

Choose a date and location to book your seat

No.

Date

Days

Location

Fees

Enrollment

27 - 29 Jul 2025

3 Days

Riyadh, KSA

$3,250

22 - 24 Sep 2025

3 Days

Dubai, UAE

$3,250

17 - 19 Nov 2025

3 Days

Online, Virtual

$1,550

Introduction

Training course introducion / brief

This course equips data scientists, ML engineers, and analysts with practical skills to accelerate every stage of the data science workflow using RAPIDS libraries , integrating seamlessly with familiar Python tools like pandas, scikit-learn, and Dask.

By the end of the course, participants will be able to:

Understand the performance limitations of traditional CPU-bound data workflows.
Leverage GPU-accelerated tools in the RAPIDS ecosystem.
Perform data loading, cleaning, and transformation at scale using cuDF.
Visualize large datasets efficiently with GPU-friendly tools.
Train and evaluate machine learning models using cuML.
Integrate RAPIDS with Dask for distributed workflows.
Compare performance metrics (CPU vs. GPU) for common tasks.

Data Scientists & Machine Learning Engineers

Data Analysts working with large-scale datasets

AI/ML Researchers

Python Developers exploring high-performance analytics

Professionals preparing for GPU-based or accelerated ML pipelines

Module 1: Introduction to Accelerated Data Science

Evolution of data science workflows
Challenges with traditional CPU-based processing
GPU computing basics: CUDA, memory, parallelism
Overview of the RAPIDS ecosystem:
- cuDF, cuML, cuGraph, cuSpatial, Dask, cuXfilter
Installation and environment setup (local or via cloud/JupyterHub)

Module 2: GPU-Accelerated Data Loading & Preprocessing

Introduction to cuDF: A pandas-like DataFrame on GPU
Loading data into GPU memory (CSV, Parquet, etc.)
Common data manipulation tasks with cuDF:
- Filtering, sorting, joins, groupby, one-hot encoding
Handling missing data and data types
Performance benchmarks: cuDF vs. pandas

Module 3: Data Exploration and Visualization at Scale

Visualizing large datasets on GPU
Integration with cuXfilter and Plotly Dash
Creating interactive dashboards
Exploratory Data Analysis (EDA) on GPU
Correlation analysis and summary statistics using RAPIDS

Module 4: Machine Learning with cuML

Overview of cuML and supported algorithms
Supervised learning: Linear Regression, Random Forest, KNN
Unsupervised learning: PCA, K-Means
Model training, validation, and performance evaluation
Integration with scikit-learn API for seamless adoption
Hands-on: Compare training times and model accuracy (cuML vs. scikit-learn)

Module 5: Scaling Workflows with Dask and RAPIDS

Overview of Dask for parallel computing in Python
Using Dask-cuDF for large, distributed datasets
Scaling machine learning with Dask-cuML
Scheduling, memory management, and cluster setup
Hands-on: Distributed training of models with Dask + RAPIDS

Module 6: Graph Analytics with cuGraph (Optional Advanced Module)

Introduction to cuGraph for graph-based data science
Graph construction from structured data
GPU-accelerated algorithms: PageRank, BFS, Connected Components
Use cases: fraud detection, recommendation systems

Module 7: Best Practices and Optimization Techniques

Profiling and debugging GPU workflows
Memory optimization strategies
Integration with cloud platforms (AWS, GCP, Azure GPU instances)
Combining RAPIDS with deep learning frameworks (e.g., PyTorch, TensorFlow)
Case studies

Boost certificate of completion

BOOST's Professional Attendance Certificate “BPAC” is always given to the delegates after completing the training course, and depends on their attendance of the program at a rate of no less than 80%, besides their active participation and engagement during the program sessions.

ENDORSED EDUCATION PROVIDER

Flexible deadlines

Customized dates accordance to your schedule

Shareable Certificate

Earn certificate upon completion

COURSE METHODOLOGY

Our Training programs are implemented by combining the participants' academic knowledge and practical practice (30% theoretical / 70% practical activities).

At The end of the training program, Participants are involved in practical workshop to show their skills in applying what they were trained for. A detailed report is submitted to each participant and the training department in the organization on the results of the participant's performance and the return on training. Our programs focus on exercises, case studies, and individual and group presentations.

Download Company profile

Download Calendar

Trending Courses

The most bespoke and flexible training courses

Open courses table

Apr

- 05 -
Days

IOSH Managing and Working Safely

Abu Dhabi, UAE

TRAINING COURSE: Accelerating End-to-End Data Science Workflow