In this episode, we dive into the fascinating world of scientific computing with Mark Kittisopikul, who has a background in computational biology. He shares insight into his career journey. And the critical role of scientific computing and modern research. Scientific computing involves leveraging computers to analyze interpret experimental data, bridging the gap between theoretical models and real world outcomes. Despite his non-computer science background, Mark helps his colleagues in biology. And other fields with computational tasks emphasizing the importance of data interpretation. He takes us through his journey, experimenting with various programming languages before discovering Julia. Which he now prefers for sufficiency in ease of use, especially in scientific applications.



Mark illustrates the benefits of Julia with the specific challenge he faced managing data from 10 cameras in a very specific microscope. Showcasing how Julia provided the solution by being able to parse gigabytes of data per second. We also explored the vibrant community around Julia and the upcoming JuliaCon conference highlighting the inclusive nature of the community and the opportunities for calibration and growth. So join us as we uncover the pivotal role of scientific computing in modern research and the advantages of using Julia for computational tasks in biology and beyond.

Send us a Text Message.

Support the Show.

In this episode, Albert recounts his journey from Nakhodka Russia to the CEO of a Dutch company Lazy Dynamics. He describes his academic trajectory from studying in St. Petersburg. To earning scholarship and master programs in Kyoto, Japan. There he focused on , developing driving aids for elderly drivers, but face challenges with system performances, leading him to pursue a PhD in Bayesian Inference. Albert explains Bayesian inference as a method for updating beliefs, about uncertain quantities based on new evidence. He discusses its applications and addressing uncertainty in complex systems like personalized. Just hearing it, the conversation touches on the differences between patient AI and reinforcement learning, I'll but also introduces RxInfer and for an open source toolbox programmed in Julia designed to automate Bayesian Inference through reactive message passing. He emphasizes RxInfer and its efficiency in handling computational resources by processing information only when necessary.

Send us a Text Message.

Support the Show.

Send us a Text Message.

This episode (recorded late 2023) of Inspiring Computing features Nikki, a tech lead in machine learning at a company called Adyen. She discusses her journey into AI and our role at adjunct. Nikki initially studied econometrics, but found it lacking in practical application. She then delved into programming, building apps and websites, and eventually combined her love for mathematics and programming.

She worked at KLM as a software engineer, primarily on C++ plus before transitioning to ING. Was she, when she began working on extensively with Python and data related projects. At Adyen Nikki explains that the company facilitates payment processes for various businesses, ensuring integration with different payment methods for companies like streaming services and eclipse Cromer's platforms. She elaborates on the, behind the scenes process of payments, including the risk checks authentication. Emphasizing agent's role as a payment. Gateway and its banking license, which allows for same day payment processing.

Nikki discusses the complexity of payment optimization due to different messaging protocols and the rules across banks, particularly. With visa and MasterCard edge and maintain standards and smooth transition of these protocols. Leveraging machine learning models, trained on past data to adapt, to changes and ensures seamless payment processing across various banks and regions. They experiment different models per region, per group, and companies to optimize this performance.

Support the Show.

In this podcast episode, we delve into the intricacies of power markets and energy forecasting with Tom Lemmens who has firsthand experience in the field. Starting his career at an energy company, our guest explains the complexities of short-term power markets, focusing on generation forecasting for wind and solar power, as well as price forecasting.

We learn about the crucial role of forecasting prices as a proxy for balancing the grid, and the importance of portfolio optimization in maximizing asset value. After transitioning from a data science consultant back to the energy sector, our guest became one of the early joiners at Dexter Energy, a company providing generation forecasting and trade optimization services.

Dexter Energy specializes in forecasting solar and wind power generation, along with short-term power prices, to help companies make informed trade strategies and optimize their assets. The guest highlights the significance of utilizing Python in their work and explains the process of translating data into expected power output using machine learning models.

Moreover, we explore the challenges and rapid changes in the energy transition, particularly in regions with increasing adoption of renewable energy sources like solar panels. Tom shares insights into the continuous evolution of their models and the technology stack used at Dexter Energy, including Python, Google Cloud, Airflow, and various databases.

Finally, we uncover the data sources for weather data, essential for accurate forecasting, and the iterative process of determining model usefulness through backtesting. This episode provides a comprehensive overview of the dynamic energy market and the vital role of data-driven solutions in optimizing energy trading strategies.

Send us a Text Message.

Support the Show.

Send us a Text Message.

In this episode of Inspiring Computing we talked to Ritchie Vink was the author of polars. Polars is a Python package written in rust and anyone who is ever used pandas will for sure find this episode interesting, as polars has incredible speedups compared to pandas. But in order to do so, it is interesting to hear from , Ritchie, where he got his inspiration from, how did he go about creating such a powerful package and where is he actually going in the future?

And this episode, he talks about some of the optimizations that he implemented.

Support the Show.

Send us a Text Message.

In this episode, we talk to Thomas who works at a company called Prodrive Technologies. We explore how the company started with these AI platform and what it meant for the company to go from simple proof of concepts of applying machine learning or deep learning techniques to actually adding business value.

We listen to the journey that Thomas and his team had to take to understand, and then reveal all the common gotches around implementing MLOps. If you're interested in understanding the tools, the challenges, the story, the motivation, the obstacles that pro drivers overcome in the recent few years in this journey.

During this episode we mention a few neat packages:
- MLFlow
- PyTorch
- TensorFlow
- Django
- anomalib

Support the Show.

Send us a Text Message.

In this episode, we talk to Pedro Holanda who explains the motivation of why it's so useful to create a company around helping data scientists, use databases. In other words, people tend to use Pandas and CSV files a lot. And this sparked his curiosity alongside a few co-founder, of why is there not a better way of using databases for data scientists?

We explore how duckDB achieved getting over 2 million downloads per month and how the secret source of making things as easy contributed to that success.

We explore some of the technical differences between SQLite and duckDB, but also a bit of the technicalities of optimising storage memory and adaptation, talking of how to develop extensions and effectively. How does the DuckDB labs as a company got going and where they're actually going in the future.

Support the Show.

Send us a Text Message.

In this episode, we explore Pluto. It is a Julia package, which enables people to ride Pluto notebooks. These notebooks are used by hundreds of thousands of people around the world to help explore the world around them. We talked to Fons to understand how he and some of his co-creators started this project, what motivated them to create yet another notebook framework and what are the key differences towards other notebooks, such as Jupiter, MATLAB Live editor, and maybe a little bit of Observable.

In addition to this, we will also explore how Pluto's are used in education with the Gerhard's experience as a teacher and a community manager, as the team collects a wide variety of use cases of where and how Pluto is used and how it helped people explore the world around. But not only do we understand and explore how it got going, how it's used today, we also take a sneak preview of what's to come in the future and understand how the community and the team make their decisions of what are the new features to be implemented and why. And it all revolves around some of their guiding principles that the team has put in place.

Useful links that are referred in the episode

• https://featured.plutojl.org
• https://github.com/fonsp/Pluto.jl
• https://julialang.zulipchat.com/

Support the Show.