Active Theory’s Post

Hey everyone, I’m excited to share that I’ve just deployed my portfolio website, which I’m calling my “Hot Page,” using a stack of Next.js 14, TypeScript, Tailwind CSS, Langchain, and Vercel AI SDK! 🧑💻 Instead of the user going through the hassle of searching through pages for answers, you can easily ask the custom AI Chatbot! It includes features like: • Handling follow-up questions 🙋♂️ • Providing external links 🔗 • Finding specific information on pages 📑 • Powered by the latest GPT-4o-mini model from OpenAI 🚀 This has been one of the most complex side projects I’ve undertaken and I learned a lot, especially as it was my first time integrating AI and designing a visually appealing website that is accessible across all screens. I look forward to refining and expanding it, whether it’s fine-tuning the model, exploring new models, or adding more data as my career progresses! Check it out ⬇️: https://lnkd.in/ezdS_gV4 #portfolio #webdev #webapp #ai #llms

Now write faster css with this ai tool 🔁🧑💻

TIC-TAC-TOE#2 - I used REACT for dynamic UI and SASS for styling, with the Minimax algorithm powering the Computer AI. I have faced a few challenges in the AI integration, specially in the recursive calls that Minimax has to do, but managed to implement it and learn a lot from it, but keep losing to my own AI 🙄 😅 Give it a go and let me know if you can beat the AI: https://lnkd.in/etii9Bsz #ReactJS #SASS #TechChallenges

Chances are high that you are using Generative AI or contemplating or reading about it. If you are generating code with Gen AI, I have an critical question for you:   What code is your AI generating?   Let me elaborate. LLMs (which is what AI is these days) are in very simple terms guessing the next token with an optimizing function. Lets do a small thought experiment with a small example software requirement - "Display a red circle of 100px radius with a blue border of 1px". We can use it to generate code at multiple levels of software and hardware stack to achieve the same goal: 1. Generate assembly code with instructions to write directly the image to be displayed on the visual display unit. 2. Generate code in C which has graphics primitives of the operating system to draw the circle 3. Generate code in HTML with SVG image for a circle 4. Generate code in Javascript which uses a library like svg.js 5. Generate code for a Vue or React component to take a color, border color and radius as parameters and draw the circle   All of the above approaches are correct and each comes with tradeoffs and limitations. Choosing the right level of abstraction is the key based on the current need and your own prediction on how much flexibility do you need for future changes or customization.   If we extend the requirements to model a 100,000 balls colliding with each other to simulate particle interactions with basic physics, which abstraction would you pick?   To quote Edsger W. Dijkstra - “The purpose of abstraction is not to be vague, but to create a new semantic level in which one can be absolutely precise.”   Are you picking the right level of abstraction when you use LLMs to generate code. Is your AI able to generate the code precisely needed for your problem statement?   --- We at kis.ai are working on this problem space, where we are building right abstraction on which LLMs can generate complete, production-ready, enterprise applications. #genai #aicode #codegeneration

For web devs , It is easy to use any open source AI model and utilize the famous transformers library ( Object detection , Text to image , Sentiment analysis , etc ...) Provided that the model must be in ONNX format to be able to used in browser enviroment. This is a great tutorial . https://lnkd.in/dFkKwaYK If you built a model from scratch , Like my case , Then follow steps in link below to convert your PyTorch model to this format. https://lnkd.in/dJWeaTp9

Project Update: FashionGAN, a Generative Adversarial Network (GAN) that generates new fashion images using the Fashion MNIST dataset. Features: -🖼️ Interactive Image Generation -🔄 Continuous Generation via Flask Backend -🚀 GPU Acceleration with Kaggle Environment Built with: -TensorFlow 🧠 -Flask 🔧 -HTML+CSS+JS 💻 FashionGAN explores the intersection of AI and fashion. Check out the repository to learn more. Thanks to: Arqam Siddiqi, Rayan Ali, and Amna Shah for their help with GPU usage and Kaggle setup. Next steps: -🔬Integrating bigger and more advanced deep learning models with web development. Repository: (https://lnkd.in/eCVB5mDC) #AI #GAN #FashionMNIST #TensorFlow #Flask #GenerativeArt #MachineLearning

Revolutionizing Real-Time Alerts With AI, NATS, and Streamlit https://lnkd.in/eTsTQfhk Imagine you have an AI-powered personal alerting chat assistant that interacts using up-to-date data. Whether it’s a big move in the stock market that affects your investments, any significant change on your shared SharePoint documents, or discounts on Amazon you were waiting for, the application is designed to keep you informed and alert you about any significant changes based on the criteria you set in advance using your natural language. In this post, we will learn how to build a full-stack event-driven weather alert chat application in Python using pretty cool tools: Streamlit, NATS, and OpenAI. The app can collect real-time weather information, understand your criteria for alerts using AI, and deliver these alerts to the user interface.

Vercel AI is revolutionizing the way AI-powered applications are built and deployed! It has some cool features such as, 1) Type a prompt to create a user interface for your web applications 2) AI SDK that makes it easy to integrate OpenAI, LangChain, and Hugging Face into React 3) Vercel AI Playground: Compare the latest AI models side by side in an interactive environment In the comments, I've given a link to an online free webinar on Vercel AI. 👇🏼 Date: Apr 22 (Monday) You'll learn: ✅ How to use the Vercel AI SDK with Generative UI ✅ How to build AI-powered applications with ease ✅ How to create dynamic user interfaces Share with folks who love Javascript and AI. #aitraining #ai #vercel

I am so proud of our Tech Research team. Check this out: Milos is introducing our Tech Research project: the development of an autonomous web explorer built on OpenAI's GPT-4 model. It functions as a highly intelligent web scraper that can perform more complex interactions, such as clicking buttons or filling out forms. This means it can access many states invisible to regular scrapers, such as a login form with an incorrect password message. It also understands the content of the pages visited and the meaning of the actions performed. The output of an AI web explorer is a simplified version of a state machine for a given website, which can be visualized as a diagram of the states visited and the actions taken to reach them. An automated web explorer can be used to create a knowledge base for designing test cases and automating their execution. 3 minutes that are definitely worth it - https://lnkd.in/eMQGPYHQ You can also follow the project in our profiq GitHub repository - https://lnkd.in/ejtUaMbD. #AI #ArtificialIntelligence #WebScraping #Automation #MachineLearning #TechResearch #Innovation #GPT4 #OpenAI #SoftwareDevelopment #AutonomousSystems #profiq #TechInnovation

Fascinating video on how Duet AI is redefining developer workflow! 🚀 https://lnkd.in/dhH58hZ8 Duet AI helps developers build web apps faster and more efficiently. The video showcases creating a simple weather app using Duet AI's code generation and task completion capabilities. ☀️ The developer built a Python Flask app with Duet AI, integrating the weather.ai.com API. He then created HTML templates for a user-friendly interface with a city input form and a results page displaying current temperature and conditions. 🌡️ All this in under 10 minutes! Duet AI is a game-changer for developers, saving them valuable time and effort. 🙌 #DuetAI #DeveloperTools #WebDev #AI #Productivity Feel free to add your own insights or questions about Duet AI in the comments! 👇