Eli Asare @mlpythoneli

David Fowler ( @davidfowl ) went from an intern to a Distinguished engineer at Microsoft. That’s 11 different promotions all at the same company. I interviewed him for everything he learned by going through that process. We discussed: • His promotions to Principal and…

⏳ Can AI truly grasp time? Meet RTQA — a framework that tackles temporal reasoning by: 🔹 Breaking down complex questions into sub-steps 🔹 Plug-and-play with any LLM (no retraining!) 🔹 Aggregating answers to cut errors & boost reliability 🚀 New SOTA results on MultiTQ &…

OpenAI just took a swing at one of AI’s biggest pain points: JOBS. By 2030 it plans to certify 10 million Americans inside ChatGPT and funnel them to employers via a new AI-powered jobs board. Early backers include Walmart, BCG, John Deere and Indeed. If the model works, AI…

keyboard_arrow_left Previous keyboard_arrow_right Next

Can AI agents reliably navigate the web? Does the choice of agent scaffold affect web browsing ability? To answer these questions, we added Online Mind2Web, a web browsing benchmark, to the Holistic Agent Leaderboard (HAL). We evaluated 9 models (including GPT-5 and Sonnet 4)…

🚨BREAKING: The wait is over. Creatify just dropped Aurora... the most lifelike AI avatar ever. One image → talking avatar. True emotions. Real expressions. Synthetic media will never be the same. Here are 6 jaw-dropping demos: 👇

Everyone thinks moving data to the cloud is simple. It's not. Here's what really happens in a cloud migration (based on my 8+ years experience):

Understanding Event-Driven Architectures: Kafka, Outbox Pattern, and Exactly-Once Guarantees - #Java Code Geeks javacodegeeks.com/2025/09/unders…

WATCH: @DigEconLab Director @erikbryn and @Stanford Professor Emeritus of CS @yshoham joined @nxthompson in a conversation about moving AI beyond hype to real impact, covering everything from pilot-to-production challenges to the future of human-AI collaboration.↘️

AI21 Labs @AI21Labs

4 days ago

WATCH: @DigEconLab Director @erikbryn and @Stanford Professor Emeritus of CS @yshoham joined @nxthompson in a conversation about moving AI beyond hype to real impact, covering everything from pilot-to-production challenges to the future of human-AI collaboration.↘️

3 9 25 59K 12

Download Video

Python vs. Excel Functions 🔹 What They Are Best At Excel Functions → Quick math, pivot tables, business reports. Python Functions → Automation, big data, machine learning. 🔹 Strengths Excel ✅ Easy to learn ✅ Great for small data & visualization ❌ Weak for automation &…

FNet (2022): What if we replace self attention with Fast Fourier Transform? - instead of letting each token ‘vote’ on what matters , mix info across the entire seq using a 2D FFT - cheaper (O(NlogN)), efficient, edge-friendly this was believed to make llms fast and efficient…

keyboard_arrow_left Previous keyboard_arrow_right Next

📣 The PyData Seattle 2025 CFP & Labor Day Flash Sale will BOTH close in TWO DAYS, Monday, Sept. 1! hubs.la/Q03GfPb_0 Get 25% off all tickets except student tickets, and submit your speaker materials by Sept. 1 for a shot at being a part of this fabulous schedule! 📣…

Many of you might have expected GPT-5 to be a huge improvement, but the results have let some people down—the gains are modest. Coding accuracy on the SWE benchmark is roughly 60%. So what happens when even the best model introduces bugs into your software? What’s the best way to…

A new report from Epoch AI confirms that GPT-5 delivers massive performance gains over GPT-4 across major benchmarks, just as GPT-4 did over GPT-3. While the release felt quieter than GPT-4’s thunderclap debut, the numbers tell a different story: 🧠 +67% on HumanEval 📚 +80%…

Epoch AI @EpochAIResearch

a week ago

A new report from Epoch AI confirms that GPT-5 delivers massive performance gains over GPT-4 across major benchmarks, just as GPT-4 did over GPT-3. While the release felt quieter than GPT-4’s thunderclap debut, the numbers tell a different story: 🧠 +67% on HumanEval 📚 +80%… https://t.co/saGsCdFQia

20 89 537 146K 138

Download Image

🚨 Excited to share new work on LLMs and loopholes, accepted to #EMNLP2025 main! When models are faced with conflicting goals and ambiguous instructions that would let them exploit a loophole, many of the strongest models (Qwen, GPT4o, Claude, Gemini) do. This is a new risk and…

“YouTube secretly used AI to edit people's videos. The results could bend reality” We live in the amnesia age where nothing from the past will be trusted— as the originals are destroyed and the artificial and synthetic becomes the fact. bbc.com/future/article…

DINO-v3 has a single high-magnitude channel on its residual pathway, channel 416. Turning off this single channel affects DINO's entire output by 50-80%. For context, turning off a random channel has an effect of less than one percent. The model builds up channel 416 in its last…

i found a very cool algorithm: poisson disc sampling it lets you place objects in a random but uniform & natural looking way it's surprising how often this algorithm shows up where you'd expect plain old .random() to work just fine it's widely used in gamedev for procedural…

Why data engineers avoid Python pandas for large-scale data work: - Pandas operates in-memory (on a single machine) - It crashes with datasets larger than RAM (common in production) - Most pandas operations use only one CPU core - Modern data processing requires distributed…

Europe's new AI plan is ambitious - now comes the hard part… By Frederik Riskær Pedersen @FRP_ET via @ForbesTechCncl forbes.com/councils/forbe… cc @BetaMoroney @Nicochan33 @enilev @mvollmer1 @mikeflache @antgrasso @FrRonconi @ramonvidall @baski_LA @AkwyZ @Khulood_Almani…

How was your first job interview experience as a Data professional? How prepared were you? Did you scale it? Did you get the job? What helped you scale through if you did? Any advice for someone anticipating their first role? “Let’s talk about it”