PyData Amsterdam

📣 Join our upcoming meetup co-organized with Snowflake ❄️ on November 7 at the Snowflake Amsterdam Office. Get ready to dive into two powerhouse talks on 𝐁𝐞𝐲𝐨𝐧𝐝 𝐭𝐡𝐞 𝐃𝐚𝐭𝐚𝐛𝐚𝐬𝐞: 𝐀𝐫𝐜𝐡𝐢𝐭𝐞𝐜𝐭𝐢𝐧𝐠 𝐒𝐜𝐚𝐥𝐚𝐛𝐥𝐞 𝐃𝐚𝐭𝐚 𝐏𝐥𝐚𝐭𝐟𝐨𝐫𝐦𝐬 𝐟𝐨𝐫 𝐭𝐡𝐞 𝐂𝐥𝐨𝐮𝐝. First up, the Booking.com Data DevEx crew İpek Özlem & Dimitar Nedev will break down their slick integration of PySpark, Snowflake, and Iceberg, moving massive amounts of data daily while keeping things fast and reliable. In the second talk, Peter Lem will detail the Actuals team's strategic migration from MySQL to Snowflake, addressing technical challenges and showcasing performance improvements in a multi-tenant SaaS environment. Attendees will gain practical lessons and innovative strategies for large-scale data platform development and migration. We have limited spots available, so, be quick to secure your spots 🔗 https://lnkd.in/eHn9qZEw. Looking forward to meeting you in-person soon!

My PyData Amsterdam talk is out! The title is "Run a benchmark they said, it will be fun they said". The talk is, as the name implies, about a benchmark. Enjoy!

📣 Next Meetup on November 7 We’re excited to co-organize our upcoming meetup with Snowflake ❄️ at Snowflake's Amsterdam office! The program will be announced very 🔜, so stay tuned to our channels and turn on your Meetup & LinkedIn notifications! And don’t forget to save the date in your calendar!

Hi all, as you may remember, in September I gave a talk at pydata on uncovering successful startups at their early stage. If you are interested in the talk and you did not have the chance to come in person to listen to it, there is the talk online on youtube! Reach out if interested in having a chat! Thanks to PyData Amsterdam to organizing such a rich event! https://lnkd.in/dbWxKNU2 #PyData2024 #PyDataAmsterdam #startup #startupsuccess #networktheory #Innovation

The PyData Amsterdam sessions have just been uploaded to YouTube. The one below is one that really stuck with me, specifically the part about the "hourglass" of dependencies.

Last month at PyData Amsterdam, I gave a talk about experimenting with new technology to reinvent how we consume news. Along with two engineers, we created Jounai - AI Nieuws: an autonomous, AI-driven news platform that was built just for fun and learning about new tech stacks (we built the backend in Java and used a lot of ML APIs). Key takeaways from the talk and this project: - ✅ Staying factual with LLM's is (mostly) possible! It is possible to make LLMs stay factual when reporting the news using old-school ML approaches and proper software engineering practices. - ❌ Hiccups happen! Mistakes in the data pipeline caused the website to declare world peace while this unfortunately was not the news. - 🕵️♂️ It is difficult to know what content on the web can be used: We recently learned news websites often hide a PDF with terms and conditions about their RSS feeds somewhere which is easy to miss. Since the platform is always open about the used sources, we luckily have been notified when we missed something and made changes. - ⚽️ Experimenting and learning are essential when you work in AI: It is very professional to play with new technology to push the limits of what is possible, as this allows you to learn a lot and innovate quickly. - 🥜 Cost effective: The platform now runs for less than the equivalent of 50 pots of peanut butter a month and barely needs human supervision. 🎥 Curious? The full talk is now available on YouTube and the platform is also available online and on Spotify.

I was asked to share the slide deck of my recent talk in PyData Amsterdam on practical challenges of Causal Inference. The video and the voiceover is out https://lnkd.in/e4879D_f

🤩 Thanks to everyone who joined our meetup co-hosted with JetBrains last Thursday! It was wonderful to reconnect after our September conference. We hope you enjoyed the talks and had some insightful conversations. Below is a summary of the talks & key takeaways: Jesse Koreman and Kally Chung’s Talk: 𝐌𝐏𝐋𝐗 - 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐃𝐫𝐢𝐯𝐞𝐧 𝐑𝐢𝐬𝐤 𝐌𝐚𝐧𝐚𝐠𝐞𝐦𝐞𝐧𝐭 focused on three core issues with the current system: ✦ Timing Mismatch: Inaccurate timing in risk assessments. ✦ Inaccurate Assumptions: Flawed risk assumptions raised concerns about insufficient deposit holdings. ✦ Merchant Dissatisfaction: Inaccurate assessments strained merchant relationships, especially regarding deposit requirements. Using a data-driven approach with survival analysis and nonparametric curve fitting, Jessie and Kally’s team achieved better accuracy, reduced deposit requirements, and improved merchant satisfaction, demonstrating MPLX’s strong impact. Bas Stinenbosch's talk 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐞 𝐲𝐨𝐮𝐫 𝐦𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠 𝐞𝐱𝐩𝐞𝐧𝐬𝐞𝐬: 𝐀𝐧 𝐨𝐯𝐞𝐫𝐯𝐢𝐞𝐰 𝐨𝐟 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭 𝐦𝐞𝐭𝐡𝐨𝐝𝐬 discussed the evolution of marketing mix models, from linear regression to Bayesian models, to optimize spend across channels. Challenges include limited weekly data points and privacy regulations, complicating multi-channel attribution. Bayesian modeling, by incorporating prior knowledge, offers actionable insights with limited data. Bas recommends combining marketing mix modeling, multi-touch attribution, and A/B testing for optimal marketing strategies. Jodie Burchell, PhD's talk 𝐋𝐢𝐞𝐬, 𝐃𝐚𝐦𝐧𝐞𝐝 𝐋𝐢𝐞𝐬, 𝐚𝐧𝐝 𝐋𝐚𝐫𝐠𝐞 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥𝐬 explored why hallucinations in LLMs happen and ways to mitigate them. LLMs, powered by vast datasets and transformer architecture, can produce impressive but sometimes misleading content. Hallucinations fall into two categories: faithfulness errors, where the model strays from the source, and factual errors, where it confidently shares inaccurate "knowledge." Solutions include prompt-tuning, self-checking, and grounding responses with live data. While hallucinations may persist, understanding and reducing their risks ensures LLMs are applied responsibly. 𝐁𝐨𝐧𝐮𝐬 𝐟𝐫𝐨𝐦 𝐉𝐨𝐝𝐢𝐞 For those in the small group chat after the talk, you already have these 😉 ! Sharing with the larger community: ✦ Another talk by Jodie: Mirror, Mirror: LLMs and the Illusion of Humanity Watch: https://lnkd.in/eZzVFC5y ✦ On the measure of intelligence https://lnkd.in/fcWYgwr ✦ Levels of AGI for Operationalizing Progress on the Path to AGI https://lnkd.in/dBNwvCjZ 𝐖𝐡𝐚𝐭’𝐬 𝐍𝐞𝐱𝐭? 🗓️ Mark Your Calendars! Our next meetup, co-hosted with Snowflake❄️, will be held at Snowflake’s office on November 7 in Amsterdam Zuidas. The program will be announced very soon, so keep an 👀 on our channel for updates!

Last month at PyData Amsterdam, I gave a talk about experimenting with new technology to reinvent how we consume news. Along with two engineers, we created Jounai - AI Nieuws: an autonomous, AI-driven news platform that was built just for fun and learning about new tech stacks (we built the backend in Java and used a lot of ML APIs). Key takeaways from the talk and this project: - ✅ Staying factual with LLM's is (mostly) possible! It is possible to make LLMs stay factual when reporting the news using old-school ML approaches and proper software engineering practices. - ❌ Hiccups happen! Mistakes in the data pipeline caused the website to declare world peace while this unfortunately was not the news. - 🕵️♂️ It is difficult to know what content on the web can be used: We recently learned news websites often hide a PDF with terms and conditions about their RSS feeds somewhere which is easy to miss. Since the platform is always open about the used sources, we luckily have been notified when we missed something and made changes. - ⚽️ Experimenting and learning are essential when you work in AI: It is very professional to play with new technology to push the limits of what is possible, as this allows you to learn a lot and innovate quickly. - 🥜 Cost effective: The platform now runs for less than the equivalent of 50 pots of peanut butter a month and barely needs human supervision. 🎥 Curious? The full talk is now available on YouTube and the platform is also available online and on Spotify.