🚀 CodeCurrent LLC Is Now Accepting New Projects & Contract Opportunities I’m excited to announce that CodeCurrent LLC is officially open to new AI, software development, and data analytics engagements as we continue expanding our capabilities across industries. At CodeCurrent, we help teams turn complex data and ideas into real, scalable solutions. Our core specialties include: 🔹 Machine Learning & Predictive Modeling – Forecasting, classification, recommendation engines – End-to-end ML pipelines ready for deployment – Advanced feature engineering and model tuning 🔹 Custom Software Development – Full-stack applications, dashboards, internal tools – API development, automation scripts, and system integrations – MVP builds for startups and innovation teams 🔹 Data Analytics & Visualization – Interactive dashboards (Metabase, Power BI, custom React) – Strategic insights driven by clean, structured data – Visual storytelling for decision-makers 🔹 Technical Strategy & AI Adoption – Identifying high-ROI automation opportunities – Integrating AI into existing workflows and products – Architecture guidance for scalable systems If your organization is exploring AI, automation, custom software, or data-driven strategy, we’d love to collaborate. 📩 DM me or reach out directly to discuss how CodeCurrent LLC can support your next initiative. — Mohamed Naji Founder, CodeCurrent LLC

Gemini 3.0 now tops the IQ benchmark with 130, outperforming all other LLMs.

This is only the beginning!!👀

ML researchers just built a new ensemble technique. It outperforms XGBoost, CatBoost, and LightGBM. For years, gradient boosting has been the go-to for tabular learning. Not anymore. TabM is a parameter-efficient ensemble that gives you: - The speed of an MLP - The accuracy of GBDT Here's how it works: In tabular ML, we've always had to choose between speed and accuracy. MLPs are fast but underperform. Deep ensembles are accurate but bloated. Transformers are powerful but impractical for most tables. TabM solves this with a simple insight: (refer the image below as you read ahead) Instead of training 32 separate MLPs, it uses one shared model with a lightweight adapter. This small tweak gives you the benefits of ensembling without the cost of training multiple networks. The results: Against 15 models and 46 datasets, TabM ranked 1.7 on average—ahead of XGBoost, CatBoost, and LightGBM. Complex models like FT Transformer and SAINT ranked much lower despite being more expensive to train. I've shared the research paper and benchmarks in the next tweet.

Google just dropped "Attention is all you need (V2)" This paper could solve AI's biggest problem: Catastrophic forgetting. When AI models learn something new, they tend to forget what they previously learned. Humans don't work this way, and now Google Research has a solution. Nested Learning. This is a new machine learning paradigm that treats models as a system of interconnected optimization problems running at different speeds - just like how our brain processes information. Here's why this matters: LLMs don't learn from experiences; they remain limited to what they learned during training. They can't learn or improve over time without losing previous knowledge. Nested Learning changes this by viewing the model's architecture and training algorithm as the same thing - just different "levels" of optimization. The paper introduces Hope, a proof-of-concept architecture that demonstrates this approach: ↳ Hope outperforms modern recurrent models on language modeling tasks ↳ It handles long-context memory better than state-of-the-art models ↳ It achieves this through "continuum memory systems" that update at different frequencies This is similar to how our brain manages short-term and long-term memory simultaneously. We might finally be closing the gap between AI and the human brain's ability to continually learn. I've shared link to the paper in the next tweet!

If you’re job searching or seeking to expand contractual employment within the tech industry- We, at CodeCurrent LLC- wholeheartedlyly encourage exploring opportunities with Seldom Technologies Pvt. Ltd. Through our founder’s prior collaboration with them, we can speak to their innovative approaches, professionalism, and strong team collaboration. Take a look! #TechOpportunities #TechCollaboration #TechIndustry #JobHunting linkedin.com/posts/seldom-te…

🚨Big Announcement📢 We are excited to unveil the final version of our latest project at CodeCurrent, a software application built for businesses -🥁please… BillFlow - Invoice Tracker !!! 🧑💻Designed with freelancers, contractors, and small businesses in mind, BillFlow - Invoice Tracker is a scalable, full-stack application 🟰 designed, developed & powered securely by our behind-the-scenes tech (rest assured, your data is safe!), to provide our clients with a clean, intuitive, and dynamic digital hub where you can take advantage of: 1. Streamlined invoice tracking & mangement🧾 2. Access to analytics that translate your business’ data into visual tools, unlocking valuable insights to empower confident decision-making & promote future expansion!📊 🧑💻BillFlow - Invoice Tracker system features include: 🔸 Secure Authentication: Using usernames and business emails 🔸 Email & Password Hashing: Your email(s) and password(s) are encrypted to internally store a corresponding, unreadable string of characters, securing & protecting your database from any would-be hackers/data breaches 🔸 Full Data Persistence/Storage: behind-the-scenes, the application safely rebuilds itself, including automated invoice ingestion, powering real-time analytic insights across your BillFlow hub’s Dashboards, without ever losing invoices, user accounts, or business metadata 🔸 Isolated User & Business Accounts: Employees can create individual accounts- separate, but monitored by your Admin business account - integrating data into your Dashboards 🧠💡BillFlow - Invoice Tracker was engineered with long-term flexibility in mind: - Database schema designed for future analytics integrations - Electron-ready structure for packaging into a desktop app - Multi-OS path detection (Windows, macOS, Linux) - Compatible with Metabase Dashboards - Support for encrypted business emails and medical IDs (med_id) 💵 BillFlow- Invoice Tracker💲 Real-time dashboards, automated invoice ingestion, and cross-platform distribution as a native app. Https://lnkd.in/e_kxhFJG 

gemini 3 pro • our most intelligent model yet • SOTA reasoning • 1501 Elo on LMArena • next-level vibe coding capabilities • complex multimodal understanding available now in Google AI Studio and the Gemini API

Congrats to Google on Gemini 3! Looks like a great model.

Meta will start including “AI-driven impact” as a formal part of performance evaluations from 2026, making employees’ use of AI tools a core expectation. "Meta will tie employees' performance to their "AI-driven impact" starting next year. The social media giant is making "AI-driven impact" a "core expectation" from 2026, Janelle Gale, Meta's head of people, told employees Thursday in an internal memo, which was seen by Business Insider."

Types of Databases

Introducing Nested Learning: A new ML paradigm for continual learning that views models as nested optimization problems to enhance long context processing. Our proof-of-concept model, Hope, shows improved performance in language modeling. Learn more: goo.gle/47LJrzI @GoogleAI

WSL!!!

Connect with us! linkedin.com/company/codecur…

Introducing CodeCurrent LLC! 👋 We’re an AI solutions company committed to helping our clients reach their goals & innovate with confidence by developing scalable, custom tools! 🧑‍💻 AI • Machine Learning • Software Engineering • Data-Driven Innovation