Data Sanity Talks (@datasanity) — Telegram channel posts

14 July 2026

Data Sanity Talks

Photo

Here is the first speaker for Data Sanity London Talks! 🎙 It's All Fun and Games Until You Deploy It in Production Speaker: Maria Matveeva, AI Team Lead @ AdTech Holding Maria's team built an LLM-powered onboarding agent to help advertisers launch and edit campaigns faster, and the prototype looked great. Then real users showed up. The token budget evaporated in days, innocent-looking prompts pried open surprising weaknesses, and — asked a simple question about traffic sources — the bot cheerfully explained that most of the traffic comes from adult websites. This talk covers what went wrong in production and how the system was redesigned to survive it. 🗓 When: Thursday, 23 July, 6:00 – 9:00 PM 📍 Where: Civo Tech Junction, London 🚀 Free — Register to secure your spot! In partnership with Sage. Learn more at datasanity.dev

285 · 11:49

21 July 2026

Data Sanity Talks

Photo

Meet our second speaker for the upcoming Data Sanity London Talks! 🎙 Qwen-Only Conversational AI Agents Speaker: Stas Bichenko, CEO @ Recombine AI agents capable of natural, human-level conversation have typically required closed-source, proprietary models. However, with the release of the latest Qwen models, it's now possible to build high-quality conversational systems using fully open-source, locally deployed models. Stas will share the practical journey of migrating a real conversational agent from proprietary models to an open-source stack, including the challenges faced along the way and the final architecture. 🗓 When: Thursday, 23 July, 6:00 – 9:00 PM 📍 Where: Civo Tech Junction, London 🚀 Free — Register to secure your spot! In partnership with Sage. Learn more at datasanity.dev

1 · 240 · 07:06

Photo

Here is the third speaker for the Data Sanity London Talks this Thursday! 🎙 AI-Driven Participation in Kaggle Competitions Speaker: Andrey Lukyanenko, Kaggle Grandmaster, ML Engineer @ Meta Can you hand a Kaggle competition to AI agents? Surprisingly, yes — but there are some caveats. Andrey will walk us through real competitions where agents wrote code, ran long experiments, and achieved strong results, while also making false assumptions, chasing dead ends, and optimising the wrong signals. We'll learn how multi-session memory, reusable skills, validation checks, and human judgement turn capable agents into a reliable research workflow. 🗓 When: Thursday, 23 July, 6:00 – 9:00 PM 📍 Where: Civo Tech Junction, London 🚀 Free — Register to secure your spot! In partnership with Sage. Learn more at datasanity.dev

296 · 12:38

22 July 2026

Data Sanity Talks

Photo

Data Sanity London Talks is already tomorrow — meet our last speaker! 🎙 Keeping Financial AI Agents Honest: A Product Playbook from Sage Speaker: Yasi Jabbari, AI Product Lead @ Sage Financial agents are easy to demo and hard to productise. Using the question “Why did my profit drop?” as a thread, this talk shares a practical product playbook for turning model capability into trustworthy agent behaviour. Yasi will cover how product teams can decide what evidence is needed, who owns the boundary, when an answer should be direct, cautious, or routed, and how to evaluate those decisions after launch. The lessons come from real financial-agent work, but apply to any high-trust AI product. 🗓 When: Thursday, 23 July, 6:00 – 9:00 PM 📍 Where: Civo Tech Junction, London 🚀 Free — Register to secure your spot! In partnership with Sage. Learn more at datasanity.dev

1 · 354 · 07:48

30 July 2026

Data Sanity Talks

Photo

Data Sanity Coffee in Belgrade — this Saturday 11 am ☕️ Come to chat about AI events over a cup of coffee (or any other beverage): ☕️ How to get the most out of an AI event? ☕️ Are big commercial AI events worth it? ☕️ Best/worst AI events you've attended. ☕️ Perspectives: participant vs speaker vs organizer. ☕️ Have you organized — or do you want to organize — an event? And, as usual, we'll exchange AI news and discuss recent developments. 🗓 Saturday, August 1st, 11:00 - 13:00 📍 Belgrade, Auditoria 🎟 Free to attend May coffee be with you.

19 · 1.4K · 09:06

16 August 2026

Data Sanity Talks

Photo

Photos from Data Sanity London Talks are out. It's been a pleasure seeing so many familiar faces — and plenty of new ones! See more photos on our website. This event was brought to you by: 🚀 Speakers — Maria Matveeva, Yasi Jabbari, Andrey Lukyanenko, Stas Bichenko 🚀 Data Sanity Team — Yaroslav Kopotilov, Nick Knizev, Nikita Severin, Stanislav Petrov, Tamerlan Tabolov, Katya Leonova 🚀 Partners — Sage, Civo Tech Junction We'll release recordings of the talks soon — stay tuned, and see you next time!

6 · 2.5K · 15:18

17 August 2026

Data Sanity Talks

Photo

Data Sanity Lightning Talks September 2 in Belgrade 🍁 We'll gather in Haos Terazije to listen to stories about AI products, workflows and side projects — from people who've put AI to work and are ready to share their worst and best stories. Format: 12–15 speakers, 5 minutes each, then networking over drinks and snacks. Got a story to tell or just want to listen? Come along! Sign up early: this event is free but the seats are limited. 📅 September 2, Wednesday 🕕 6:00–9:00 PM 📍Haos Community Space, Terazije 🔗 Sign up here

25 · 1.2K · 11:03

1 September 2026

Data Sanity Talks

Data Sanity Lightning Talks — tomorrow! September 2 | Belgrade We're happy to share our speaker lineup: 🎙 Ilya Zakharov — Code in Space: What Software Engineering Can Look Like in Extended Reality 🎙 Alex Prismotrov — Vibe-Coding Browser Automation with Claude (Job Hunt Tool as an Example) 🎙 Stanislav Starchenko — AI in Cold Outreach 🎙 Konstantin Bolotov — Self-Studying Chronicles in the Age of AI 🎙 Aleksei Shevkoplias — AI Agents for Life Science Tasks: An Overview 🎙 Ilya Kareltsev — Creating a Game with an AI Agent 🎙 Katherine Patchett — What AI Reads vs. What Humans See 🎙 Nikola Mitiċ — Collective Innovation. How AI Moves Innovation to the Edge of the Organization 🎙 Viktor Vialov — A Family Vacation Film: Codex, FFmpeg, and a Bit of Engineering 🎙 Konstantin Burkin — Beyond the Average: Multi-Model Routing and Smarter Metrics for High-Stakes Pricing 🎙 Dmitrii Khodorkin — From Touch to Traction: How My 3D Moon Pet Project Became My Best-Performing LinkedIn Post Join us for an evening of short talks about building with AI, creative experiments, lessons learned and the cool things people make. The event is free, but seats are limited — save yours while there’s still room. We also have 2-3 speaker slots left, so if you have a project or story to share, come and shine! 📅 Wednesday, September 2 🕕 6:00–9:00 PM 📍 Haos Community Space, Terazije 🔗 Sign up here

8 · 310 · 10:00

5 September 2026

Data Sanity Talks

Photo

Here are some photos from the Data Sanity Lightning Talks in Belgrade. It was great to see so many faces! Special thanks to our brave speakers and to Haos Space for hosting this event. 🔗 Link to all photos

1 · 296 · 14:24

16 September 2026

Data Sanity Talks

Photo

Data Sanity Coffee in Belgrade — this Saturday at 11:00 ☕ This time, we're talking about AI devices: ☕️ How do you integrate AI into a device? ☕️ AI on the edge vs. in the cloud ☕️ Top AI devices on the market ☕️ Personal data & privacy And more — expect AI news, tips and tricks, and personal stories. 🗓 Saturday, September 19, 11:00 - 13:00 📍 Belgrade, Auditoria 🎟 Free to attend May coffee be with you.

23 · 939 · 17:19

19 September 2026

Data Sanity Talks

ReplyData Sanity Coffee in Belgrade — this Saturday at 11:00 ☕ This time, we're talking about AI devices: ☕️ How do you integrate AI into a device? ☕️ AI on the edge vs. in the cloud ☕️ Top AI devices on the market ☕️ Personal data & privacy And more — expect AI news, tips and tricks, and personal stor

Data Sanity Coffee starts in 2 hours ☕️ See you in Auditoria!

1 · 145 · 07:02

26 October 2026

Data Sanity Talks

Photo

What would you hand over to AI, and what should you never hand it? AI models keep improving, but they still have real limits, and misusing them can go horribly wrong. Data Sanity will run the Lightning Talks block at Cognitive Code Conference this Wednesday. Four short talks with lessons from real projects: 💬 “Can AI create AI? Wins and failures from real agentic projects” by Yaroslav Kopotilov and Nikita Severin 💬 “Should your AI make people feel better? How helping humans is a design problem” by Elizaveta Musatova 💬 “Sales-Driven Development: keeping a foundation when every deal arrives with a new feature request” by Ivan Petrovic 💬 “Where AI serves the reader, and how we make it” by Dusan Veljovic and Nebojsa Vojvodic Catch us at 15:00! 🗓 October 28, Wednesday, 9:00 - 16:00 📍 MTS Dvorana (Central Hall), Belgrade 🔗 Full agenda: https://cognitivecode.io/agenda

2 · 161 · 15:06