Frederick Ayala

📢 As part of Data Month at Capitole, check out our latest blog post by data expert Ignacio R. where he explores the world of large language models (LLMs) and their limitations. In this thought-provoking article, Ignacio breaks down what LLMs are, how they function, and the constraints they face despite their impressive capabilities. Whether you’re a data professional or an AI enthusiast, this post offers critical insights and practical perspectives on the evolving role of LLMs in AI. 🔗 Dive into the full article here: https://lnkd.in/eMWM3y7B 👉 Stay connected with Capitole for more expert insights, industry trends, and updates. Together, let’s navigate the future of data with a clear understanding of the tools at our disposal!

27

1 comentario

In a comprehensive new guide, Eivind Kjosbakken's walks us through the inner workings of the SAM 2 model and explains how you can start leveraging its power for your own image-segmentation tasks.

12

🎮 In the gaming industry, turning dreams into reality often depends on securing the right funding. The more we know about it, the better we can enhance the efficiency and stability of our companies in this competitive market. 🚀 Want to dive deeper into investments in animation and video games? Join us at WEIRD MARKET in Valencia, Spain on Thursday, October 3rd from 11:15 CEST to 12:00 CEST. 🎤 The table will feature our CEO Sergio Reyes and other experts in the investment gaming industry. 💼 Come and learn how to attract investment and give an exciting twist to every pitch you make! #Investment #GamingIndustry #Animation #VideoGames #WeirdMarket #Valencia2024 #Entrepreneurship

19

A new #Druid30 feature highlight from Hellmar Becker: "Druid Data Cookbook: Parameterizing the IN clause" https://bit.ly/4bhpgt2

8

1 comentario

How we boosted FormelSkin’s ROI by 15% in 4 Months with ML: https://lnkd.in/dP4nfkDq We recently worked with FormelSkin, a Berlin-based health-tech startup specializing in personalized acne skincare solutions. They faced a significant marketing challenge: determining the most effective ad strategies to attract long-term subscribers instead of one-time buyers. The company's subscription model required significant upfront marketing expenses, and returns could be delayed for up to two years, raising concerns about the return on investment (ROI). To address this, we partnered with FormelSkin to develop a reliable framework for assessing the efficiency of their ad campaigns. This framework allows FormelSkin to identify and focus on campaigns that attract loyal customers. Using end-to-end analytics and machine learning, we were able to predict customer loyalty and assess the effectiveness of all ongoing ad campaigns. This way, FormelSkin could focus more on campaigns that attract loyal, long-term customers instead of investing heavily in campaigns that do not deliver value. As a result of our collaboration, FormelSkin could now: 1. Predict the long-term value of customers and adjust their marketing budget towards more effective campaigns, 2. Experience a 15% increase in their marketing ROI, which continues to improve as the average customer lifetime extends. Learn more by reading the complete case through the link above.

2

Back in November 2023 Tiago Teixeira Correia wrote an epic piece on AI in Gaming: Hype or Reality? https://lnkd.in/eDPxpT5i But in AI, 11 months is like a lifetime ago... Our upcoming conversation explores how AI has been busy reshaping game engines, value creation, competitive moats and where investors are now placing their bets in GenAI X Gaming. Register to be notified: https://lnkd.in/ek7j-EK3

3

No JVM. No problem. Redpanda frees you from time-draining #JVM tuning and fiddling with external dependencies for super simple deployments. So if you want to get straight to #streamingdata without the faff of JVMs, take Redpanda for a free spin! https://lnkd.in/gppCdNb9 #streamingdatadoodles

25

4 comentarios

Apache Arrow has become the de facto standard for high-volume data transfer. But did you know that ClickHouse can stream results in Arrow over HTTP! This makes it simple to integrate with high-performance visualization libraries like Perspective to stream real-time data. Using Apache Arrow, our latest demo efficiently transmits data from ClickHouse directly to the browser, delivering smooth visualizations in just 100 lines of code! 📊🚀 - ⚡ Arrow for efficiency: Apache Arrow minimizes the data transferred and enables smooth data streams from ClickHouse to Perspective. - ⏱️ Real-time streaming: Forex data streamed into a browser is updated every 30-50ms, showcasing ClickHouse's capability to serve hundreds of queries per second. We can even stream the entire 60GB dataset in minutes to the browser! - 🔄 Seamless integration: Built with performance in mind, Perspective handles millions of data points in the browser with real-time pivoting, filtering, and transformations. Curious? Check out the full blog to explore more or just play with the demo. https://lnkd.in/e8E_y4vS

69

𝗘𝗩𝗔 𝗠𝗮𝗹𝗮𝗴𝗮 𝗪𝗿𝗮𝗽𝘀 𝗨𝗽: 𝗗𝗮𝘆 𝟮 𝗛𝗶𝗴𝗵𝗹𝗶𝗴𝗵𝘁𝘀! Day 2 of the European Videogame Accelerator (EVA) program in Malaga was all about refining pitches, diving deeper into expert guidance, and celebrating the creativity and potential of participating studios. 𝗠𝗼𝗿𝗻𝗶𝗻𝗴 𝗥𝗲𝗰𝗮𝗽: The day kicked off with a high-stakes pitch session! Studios presented their game concepts once again, this time with the added pressure (and support) of their dedicated mentors sitting in to observe. This valuable exercise allowed studios to showcase their progress and incorporate feedback from their mentors. 𝗦𝗵𝗮𝗿𝗽𝗲𝗻𝗶𝗻𝗴 𝘁𝗵𝗲 𝗘𝗱𝗴𝗲: Following the pitches, the end of the morning focused on one-on-one mentoring sessions once more. Studios participated in another round of the "Expert Shuffle," receiving personalized advice and insights from a variety of industry veterans. This ensured a well-rounded perspective and addressed specific challenges faced by each studio. 𝗖𝗲𝗹𝗲𝗯𝗿𝗮𝘁𝗶𝗻𝗴 𝗦𝘂𝗰𝗰𝗲𝘀𝘀: As the program drew to a close, a sense of accomplishment filled the air. Studios had completed two intensive days of learning, networking, and refining their game concepts. 𝗟𝗼𝗼𝗸𝗶𝗻𝗴 𝗔𝗵𝗲𝗮𝗱: With the foundation firmly set, studios are now equipped with the knowledge, connections, and strategies to propel their games to the next level. The learnings and camaraderie fostered during EVA Malaga will undoubtedly serve them well on their development journeys. 𝗙𝗮𝗿𝗲𝘄𝗲𝗹𝗹, 𝗠𝗮𝗹𝗮𝗴𝗮! We bid a bittersweet farewell to Malaga, but the memories and connections made during these days will continue to inspire and empower participating studios. 𝗡𝗲𝘅𝘁 𝘁𝗼 𝗰𝗼𝗺𝗲 𝗶𝘀 𝘁𝗵𝗲 𝟴𝘁𝗵 𝗼𝗳 𝗠𝗮𝘆 𝘄𝗲𝗯𝗶𝗻𝗮𝗿 𝗮𝗻𝗱 𝗼𝗻𝗹𝗶𝗻𝗲 𝗪𝗼𝗿𝗸𝘀𝗵𝗼𝗽 𝟮 𝗼𝗻 𝘁𝗵𝗲 𝟮𝟴/𝟮𝟵 𝗠𝗮𝘆. 💛 #EVA is co-funded by EU Digital & Tech / Creative Europe MEDIA 💛 Thanks to our program partners Arctic Game and Polo Nacional de Contenidos Digitales de Málaga 💛 A big thanks also to our partners OFAJ_DFJW, ICEX, Kickstarter, WBS.LEGAL and Marchsreiter Communications

29

🚀 New in txtai 7.3: Streaming RAG. This feature builds off streaming LLMs which iteratively return chunks of content as a stream vs waiting for the entire generation call. Streaming is supported with Transformers, llama.cpp and LLM APIs (i.e. GPT-4, Claude, Ollama) Try it out with txtai's new RAG application. Plug in a new txtai embeddings index to use this application with your own data. Docker image: https://lnkd.in/d-fJiVXR RAG Application: https://lnkd.in/e82fddyY

5

Are you interested in integrating LLMs into production-ready applications, but you're worried about model accuracy or hallucinations? Then you're not going to want to miss our latest webinar! Fine-Tuning Enterprise RAG Knowledge Bases with Label Studio, ChatGPT, and Ragas Join HumanSignal’s Jo Booth for a hands-on workshop/webinar where he’ll show you how to build a RAG application using Label Studio, ChatGPT, LangChain, and Ragas to create, evaluate, and fine-tune an enterprise knowledge base and ensure that the content is accurate and properly aligned. https://lnkd.in/dPrJ8kbq #RAG #machinelearning #mlops #ragas #langchain #chatgpt #llm #modelevaluation

7

At #dataing, our goal is to put users in the best position to succeed by AI technology. Current dating apps often struggle to generate matches with similar interests because they rely heavily on user-inputted data, which can be incomplete or inaccurate. With our technology, we are able to analyze user interest based on photos and digital footprints to generate compatible matches. Feel free join our waitlist for early access for all features 👇 https://lnkd.in/gwfGs-Av #Love #Relationships #AI

2

📝 Release Notes: v12.4.0 📝 We’re thrilled to announce the release v12.4.0 of the AI Suite, packed with several enhancements and optimizations! 🚀 What’s new? Features ✅ Modify & generate text within CKEditor ✅ New APIs: DeepL, Anthropic ✅ Translate Content with AI, Deepl, Google Translate ✅ CKEditor available in content generation process

2

#DataPOEM life 𝗖𝗹𝗶𝗲𝗻𝘁: 𝗛𝗼𝘄 𝗹𝗼𝗻𝗴 𝘄𝗶𝗹𝗹 𝗶𝘁 𝘁𝗮𝗸𝗲 𝘁𝗼 𝗴𝗲𝘁 𝗥𝗢𝗜 𝗿𝗲𝗮𝗱𝘀 𝗼𝗳 𝗻𝗲𝘄 𝗰𝗮𝗺𝗽𝗮𝗶𝗴𝗻𝘀 𝗼𝗻 𝗡𝗲𝘁𝗳𝗹𝗶𝘅, 𝗧𝗶𝗸𝘁𝗼𝗸 𝗲𝘁𝗰 Our Answer: We can provide the ROI of new features whether it is  Netflix, TikTok Campaigns in Just 4-8 Weeks! Thanks to our Causal AI-based incremental training methodology Traditional media mix modeling (MMM) approaches often struggle to provide timely insights, leaving brands in the dark about the ROI of their investments. Waiting months to measure the impact of your new Netflix or TikTok campaigns can mean missed opportunities in today's fast-paced digital world. Harness the power of monthly ROI reads to get insights on these new media features in just 4-8 weeks. With monthly incremental model training using a Causal AI-based approach, you can quickly assess the ROI of your groundbreaking Netflix partnership or viral TikTok challenge. This empowers you to make data-driven decisions, optimize budget allocation, and scale successful campaigns in real time. No more prolonged waiting periods. Monthly ROI reads provide timely, accurate data to inform your marketing strategy and keep you ahead of the competition. Embrace the future of marketing analytics and elevate your brand. #MonthlyROIReads #NetflixCampaigns #TikTokChallenges #MarketingAnalytics #RealTimeInsights

20

🎙️ Episode: Mastering Data Science in the Gaming Industry with Anthony Generoso Pascale and.. DATA QUALITY from Miniclip🎙️ 🔍 In this episode, we uncover: ✅ Data Science Journey: Anthony's path from quality and monetization roles to becoming a data expert in the gaming industry.  ✅ Defining Roles: Clear distinctions between data engineering, data analysis, and data science, and their importance in gaming.  ✅ Building a Data Team: Essential steps for setting up a data team, choosing the right tools, and ensuring timely data integration.  ✅ Data Quality: The critical role of data quality in analytics and decision-making, with tips for thorough testing and integration.  ✅ Privacy and Analytics: How privacy changes impact data collection and analysis, and strategies for adapting to these changes.  ✅ A/B Testing: Best practices for conducting A/B tests, including hypothesis formulation, dependent variables, and comprehensive metric analysis.  ✅ Future of Data Management: Insights into how data management might evolve, with an emphasis on new tools and effective data architecture.  ✅ Practical Advice: Key recommendations for game studios on integrating data analytics early and maintaining high data quality. 📈 Ready to Elevate Your Data Science and Gaming Analytics Knowledge? 📈 Comment "LEVEL UP" below, and I'll share the podcast with you! Don’t miss this chance to learn from a leading data expert and take your game development skills to the next level! 🚀 https://lnkd.in/eYTqbkFY #DataScience #GamingIndustry #DataQuality #DataEngineering #DataAnalysis #DataTeam #ABTesting #GameDevelopment #PrivacyAndAnalytics #FutureOfData #GameAnalytics #Monetization #DataExpert #Podcast #LevelUp

2

Interested in learning how #Presto can better support ML users? Join us at #PrestoCon Day where Pedro Pedreira will give a talk on just that 👏 Register 👉 https://cvent.me/VxdxB8 Based on his experience supporting #machinelearning workloads at Meta, Pedro will present some of Meta's findings on using Presto to prepare large scale training datasets. He'll also share the team's progress on supporting these workloads and the initial results. See the full agenda at https://lnkd.in/eERCBfF6

22