Insights from PyCon US 2024: Data Storytelling and Visualization with uPlanner’s Chief Data Officer
We were delighted to connect with Sebastian Flores, our esteemed Chief Data Officer, for an exclusive insight into this year’s PyCon US conference held in Pittsburgh, Pennsylvania. Sebastian discusses the pivotal themes addressed, his sessions on data storytelling and visualization, and emphasizes the critical role of such events in fostering learning, innovation, and global networking.
Q: We heard you attended PyCon US these year. Can you share your history with this event and your experience at this year’s US event?
PyCon conferences, held annually in various countries, are key events for the Python community. They highlight local community work and foster connections. I’ve been attending PyCon events in Latin America for five years, starting with PyCon Colombia in 2020. During the pandemic, I participated in several online PyCons including Argentina, Chile, and and PyCon Latam, the regional Latin America one.
Post-pandemic, I became actively involved in the Python Chile community, helping to organize events. While organizing, the focus shifts from programming to logistics, but it’s rewarding to expand your network and meet people interested in these events. PyCon is very inclusive, welcoming both newcomers and experienced developers equally. Python’s diverse applications mean everyone is always learning something new. Attending PyCon US, the largest of these events with over 3,000 participants, was incredible. It felt like being in the ‘Hollywood of programmers,’ meeting key figures like Guido Van Rossum, the creator of Python. It was an impressive and inspiring experience, reinforcing the value of community and collaboration.
Q: Within the context of uPlanner’s technology, how does attending this type of conference inform your work and our product development?
It’s mainly about staying in tune with the latest technological advancements, new libraries, and trends observed in other companies. These events allow conversations with people from various industries and countries.
PyCon features an overwhelming number of talks—over 100 across five tracks in three days. These talks are selected from numerous submissions by different speakers, highlighting evolving trends. Five years ago, the focus was on classic Machine Learning, and how to train algorithms with large datasets. Today, trends include developing Large Language Models (LLMs) like ChatGPT, understanding their limitations and advantages, and the rise of observability—taking system monitoring to the next level by fully instrumenting systems for real-time insights. These conferences expose attendees to industry trends and company innovations, with company booths offering engaging interactions about frequently used libraries and upcoming developments. Overall, it’s a highly enriching experience.
Q: Can we dive deeper into major themes from this year that will continue to be relevant throughout 2024 and influence the upcoming year?
Regarding LLMs, these models, such as the one behind ChatGPT, enable generative capabilities like chatbot interactions. They aren’t ‘dark magic’ but are based on statistical models that generate text word by word. Several talks discussed libraries for using these models and a keynote highlighted their limitations and advantages. LLMs are powerful tools, especially for code generation, because they can then follow strict rules. They can simplify programming by generating code based on specific requirements.
It’s important to note these models don’t replace programmers; on the contrary, I believe it will enable many more people to program. The real challenge isn’t writing code but rather understanding the problem we want to solve with it. This level of metacognition is difficult. It’s easy to type away and believe you’ve solved a problem just because there’s code. The true challenge lies in interpreting what the client has communicated—what they really meant—and modeling it appropriately. This process doesn’t initially involve code; it’s about modeling, abstraction, understanding the true intent of the other party, all while navigating through data, potential gaps, coherence issues, and more. The tools available to us don’t make programmers redundant. They allow programmers to focus on the challenging aspects like abstraction, while the more straightforward, automatable tasks—like code generation—become increasingly fluid. We still need to review and not blindly trust what machines produce, but it certainly makes the process easier.
Observability, on the other hand, takes system monitoring to the next level. It involves fully instrumenting systems to provide real-time insights into their operations, both when functioning correctly and when failing. This trend is about knowing exactly what is happening in a system at any given moment, which is crucial for maintaining and improving software performance.
These themes will continue to influence the tech landscape through 2024 and beyond, offering new tools and approaches for both developers and organizations.
Q: During the conference you hosted a talk and an open space session. Can you share details about the session, your intention behind it, and what you achieved?
The talk explored data storytelling techniques and highlighted Python tools at our disposal to design effective and engaging presentations. I was pleasantly surprised by the packed room and the engaging questions that followed. It was validating to receive positive feedback and insights from attendees who resonated with the need for better storytelling and underutilized Python capabilities. Often, until tools become massive, it is only by recommendation that you get to learn about other tools available.
Additionally, I facilitated a more informal session within the open spaces format at PyCon. This session was titled ‘Data Storytelling: The Revenge of the Audience,’ where around 20 participants gathered. We discussed various challenges in data visualization, shared recommended books, and explored diverse topics spontaneously. The informal setting encouraged an equal exchange of ideas, contrasting with traditional speaker-audience dynamics. It was refreshing to see how these sessions can spark unexpected collaborations and meaningful conversations. For instance, my participation at PyCon Chile in 2021 led me to collaborate with the library Streamlit, which we now use on some of our company projects.
Q: What are your thoughts on effective data visualization?
I believe it’s similar to our discussion about LLMs, where technology advances make things mapping increasingly simple. Tools like Power BI, Tableau, and even Python libraries, alongside Excel, enable relatively straightforward graph creation. However, the ability to easily produce multiple charts for a PowerPoint presentation doesn’t necessarily add value. It’s crucial to distinguish between the volume and quality of visual production. Often, we struggle to pause and ask ourselves if we truly need to display all the data. What’s essential is distilling the information: identifying key insights, actionable items for the audience, and crafting a narrative that supports these points with relevant graphs. Often, we simply input data into an Excel spreadsheet, generate a bar chart, and consider the job done.
In reality, effective data visualization requires intentionality and care. Intentionality ensures that we provoke a specific reaction or prove a point, guiding how we structure and present data. Care involves treating visualization as an art form, taking extra moments to refine design, choose colors intentionally, and enhance aesthetic appeal. Too often, default settings in graphing software become our ending point, rather than a starting point and a baseline for improvement.
Q: What were the most significant takeaways from the event for you?
The significant interest in Data Storytelling and Data Visualization, topics that I am deeply passionate about. I often find myself discussing these subjects and sharing my enthusiasm, although I sometimes worry about repeating myself. What I find fascinating is how these fields reconcile two seemingly opposing worlds: data analysis and visualization, and the human need to consume stories.It’s gratifying to see how they converge due to our singular human mind.
I also came away with the impression that the human talent in Latin America is exceptionally strong. I came across highly relevant topics and met professionals who are undertaking impressive projects. I connected with several people on LinkedIn, and was impressed by the variety and quality of their global roles and responsibilities. Most reassuring was encountering individuals who were approachable, friendly, and humble, despite their impressive backgrounds.