2025-04-26 – 20:40-21:05 (Africa/Abidjan), Barn
Looking to make your scientific code faster and more efficient for training and prototyping? In the AI era, computing requirements and expenses are outpacing available resources. To stay ahead, efficient code is no longer a luxury - it has become a NECESSITY.
We will delve into the best practices and techniques for an optimized utilization of Python scientific libraries, including NumPy, Pandas, and SciPy highlighting common pitfalls and failure modes that can hinder performance. By going through some live demos ("what could possibly go wrong?" - North Bay Python 2024), we will learn how simple changes to our code can lead to significant optimizations, resulting in faster and more capacity-friendly solutions. We will gain insights into advanced programming concepts, such as NumPy vectorization, column-based vs row-based data frames and arrays, and many more.
Whether you're a data scientist, machine learning engineer, researcher, or software engineer, we will showcase tips and strategies that can improve the efficiency of your code and accelerate your AI projects.
No sensitive information.
Parul is a Senior Production Engineer at Meta, where she works within the Python Foundation team. Her focus is on Python's PyPI packages for internal Meta use-cases, to deliver an accelerated research-to-production AI development. She is also an early contributor to FairLearn, an open-source Python library to help assess and mitigate bias in AI systems.
Parul is also an advocate for bridging the gender gap in technology and actively mentors several aspiring technologists. Her goal is to use her expertise to inspire and support others to build a career in the field of computer science.
In her free time, Parul enjoys dancing, painting, and traveling.