This page is a long personal bookmark list of AI tools, learning material and odds and ends. The tools come first, from chat assistants and search engines to video, audio, image and slide generators, each with a line on what it does. The middle section gathers machine learning and deep learning resources, including courses, Vietnamese translations of popular books, cheat sheets and articles on topics like semi supervised learning and object detection. The last stretch holds unrelated bookmarks, such as career roadmaps and forum threads. The reader gets a single place to scan for a tool or a study path, though many entries carry no grouping.
- TheB.ai is an AI platform that gives unified access to many large language models and AI tools through a single interface. Users can chat with different AI models, generate content, and use AI for productivity and creative tasks, acting as a hub for leading AI technologies without needing to sign up for many separate services.
- Finchat.io (now Fiscal.ai) is an all-in-one AI-powered stock research platform providing financial data for global stocks, ETFs, and funds. Features include AI summaries, KPI data, personalized dashboards, and an API. Trusted by more than 350,000 investors.
- OpenFuture AI is the world's largest artificial intelligence tool directory with over 40,000 AI tools, categorized by field including productivity, development, marketing, health, e-commerce, and finance. Supports multiple languages including Vietnamese.
- Jasper AI is an AI marketing automation platform for modern marketing teams. It offers 100+ specialized AI agents, automated content workflows, a brand context hub, and a multichannel content creation toolkit for SEO, personalization, and ad campaigns.
- Adobe Podcast is an online AI-powered audio recording and editing tool. The web-based platform improves recording quality, removes background noise, and enhances voice audio, aimed at podcast creators and audio content producers.
- Blockade Labs Skybox AI is an AI tool that generates 360-degree skybox environments and panoramic worlds from text prompts. Widely used in game development, virtual reality experiences, and creative projects, it lets users create, remix, and export high-quality HDRI skyboxes for 3D engines such as Unity and Unreal Engine, replacing the costly manual skybox workflow with fast AI generation.
- SlidesAI.io is an AI presentation tool that integrates directly with Google Slides, automatically turning text into a complete slide deck with layout, content organization, and design. The pricing page offers a free Basic plan with 12 presentations per year, a Pro plan at 8.33 USD/month for students, and a Premium plan at 16.67 USD/month for professionals with unlimited presentations and document upload support, an optimal solution for anyone who needs to make slides quickly.
- Runway is a leading creative AI platform offering advanced video creation and editing tools built on machine learning models such as Gen-2 and Gen-3. It lets users create, edit, and process video content using text prompts, images, and existing footage. Widely used by filmmakers, designers, and creative professionals for AI-based video production, from generating new video to changing style and removing objects from footage.
- Fliki is an AI text-to-video platform that turns scripts, blog posts, and text content into engaging videos with realistic multilingual AI voiceover, automatic subtitles, and a rich stock media library. Users can create professional-quality videos for social media, marketing, education, and content creation without prior video editing experience, streamlining video production for content creators and marketers.
- Tome is an AI-based presentation tool with storytelling features, automatically generating narrative-style slide decks with artificial intelligence. It combines AI-generated text, images, and layouts to produce visually engaging presentations. Users just enter a topic or outline, and the platform creates a complete deck with contextual images, structured content, and a consistent visual theme, changing the traditional approach to presentation design.
- Futurepedia is a comprehensive AI tool directory and discovery platform that curates and categorizes the latest AI products and services. It helps users search, compare, and evaluate AI tools across categories such as productivity, design, writing, video, marketing, and development. The platform features user reviews, ratings, and daily updates on newly released AI tools and industry trends, an essential resource for anyone keeping up with the AI ecosystem.
- TutorAI.me is an AI learning platform that lets users learn any topic by generating custom courses with AI.
- BAI Chat (chat.theb.ai) is a free online AI chatbot platform offering access to large language models similar to ChatGPT. Its clean web interface lets you chat with AI, ask questions, generate text, translate languages, and perform many natural language processing tasks without signing up or paying, a convenient alternative to paid AI assistant services that makes advanced conversational AI accessible to everyone.
- Poe is Quora's AI chatbot platform that gives access to multiple AI models such as ChatGPT, Claude, and others in a single interface.
- Andi Search is a next-generation AI search engine that provides intelligent search results. Unlike traditional search, Andi combines AI that understands query context and answers directly instead of just listing links.
- Stockimg AI is an all-in-one AI design and content creation platform for generating logos, stock photos, illustrations, wallpapers, Instagram posts, and TikTok videos in seconds. It also offers social media management features including post editing, scheduling, and automation integrated with more than 10 platforms such as Instagram, TikTok, LinkedIn, YouTube, and X. Recognized as a leader on G2 and used by employees from Netflix, Spotify, Tesla, Shopify, and Airbnb.
- Character.AI is a platform for creating and chatting with custom AI characters, from famous characters to user-created ones.
- Texti.app was an AI writing assistant that helped users create and improve text directly in the browser, with AI text generation and editing. The service officially shut down on August 29, 2025, and the domain is now listed for sale by its parent company muze.digital. Former users can still subscribe to the AI newsletter for news and AI-related developments from the team.
- AssemblyAI is a leading AI platform for speech recognition (speech-to-text) and audio understanding. It provides a highly accurate speech-to-text API supporting 99+ languages, speaker detection, sentiment analysis, and real-time streaming.
- OneMoreAI is an AI content generation platform offering tools to create many kinds of text and image content. The app provides templates and AI workflows for marketing copy, blog posts, social media content, product descriptions, and other creative materials, designed to streamline content production for marketers, content creators, and businesses that want to use AI to create content faster and more efficiently.
- Remini AI Web is an AI photo enhancement tool that restores and sharpens blurry, old, or low-quality photos.
- Chatsonic by Writesonic is an advanced AI chatbot and conversational assistant positioned as a ChatGPT alternative with real-time data access. Built into the Writesonic platform, it enables factual content generation, answers to questions with up-to-date information, AI image generation, and voice interaction. Writesonic offers 100+ templates for articles, ads, landing pages, and more, serving marketers and businesses worldwide.
- Bing AI Chat is an AI chatbot built into Microsoft's Bing search engine, using the GPT-4 model to answer questions and generate content.
- Claude AI is Anthropic's AI chatbot, designed to be helpful, harmless, and honest, with the ability to handle long text and complex analysis.
- Explainpaper is an AI research-paper reading tool that lets you highlight hard-to-understand text in academic papers and get instant simplified explanations from a large language model. Supports 50+ languages, adjustable complexity levels, and Q&A chat.
- WriteMail.ai is an AI email writing helper that produces professional emails 87% faster, with tone control, personalization, style conversion, and support for 30+ languages. It offers new email generation and reply drafting.
- Dewatermark.ai is a free online AI watermark remover that removes watermarks, logos, text, and timestamps from images without blurring, using AI inpainting and generative fill. Supports JPG, PNG, and WEBP with batch processing of up to 50 images.
- Google AI Test Kitchen is Google's AI experimentation playground (now part of labs.google/fx) for exploring and interacting with cutting-edge creative AI tools including MusicFX, Whisk, Flow, and Project Genie.
- An article analyzing the role and limitations of Non-Maximum Suppression (NMS) in object detection, explaining what causes duplicate predictions from surrounding anchors. The author presents two necessary conditions for eliminating NMS: using a combination of classification cost and localization cost for label assignment, and applying one-to-one label assignment instead of one-to-many. It also introduces effective end-to-end object detection methods that let the model work without NMS post-processing.
- A post in the Machine Learning cơ bản Facebook group collecting comprehensive cheat sheets for Data Science beginners. It shares quick references on statistics, machine learning, Python, pandas, numpy, matplotlib, and scikit-learn, helping newcomers approach data science more systematically and effectively.
- An article on semi-supervised learning to address the shortage of labeled data. The author introduces foundational assumptions such as the Smoothness, Cluster, Low-density Separation, and Manifold Assumptions. The methods analyzed include Consistency Regularization, Pseudo Labeling, and hybrid algorithms such as MixMatch and FixMatch. It explains the mathematical basis and core idea of each method for ML problems with scarce labeled data.
- A Facebook post by AI4E (AI for Everyone) sharing educational material on artificial intelligence, from basic to advanced knowledge of machine learning, deep learning, and real-world AI applications. AI4E is a community project to spread AI knowledge to everyone in Vietnam.
- An interactive Machine Learning and Data Science learning roadmap from Delta Academy, helping learners chart a path from basic to advanced knowledge. The map visualizes the areas to study, including statistics and math, Python, machine learning algorithms, deep learning, and deployment, helping newcomers build a systematic study plan.
- The official site of the project by MLBVN Group to translate the book Hands-on Machine Learning with Scikit-Learn, Keras & TensorFlow (2nd edition) into Vietnamese. Two volumes totaling 769 pages, released as PDF.
- A Facebook post by AI4E (AI for Everyone) sharing material and content related to machine learning and AI. Covers popular ML algorithms, deep learning frameworks, and practical applications of artificial intelligence in life and work.
- A post in the Machine Learning cơ bản Facebook group discussing useful Python libraries for data science. Members share their experience using pandas, numpy, scikit-learn, matplotlib, seaborn, and many other libraries for data processing, visualization, and building effective machine learning models.
- Made With ML, now part of Anyscale, is a comprehensive education platform by Goku Mohandas that teaches developers how to design, develop, deploy, and iterate on production machine learning applications. With over 40,000 learners, it covers the whole MLOps lifecycle including data preparation, model training, experiment tracking, evaluation, serving, testing, CI/CD workflows, monitoring, and data engineering. It offers foundational courses on Python, NumPy, Pandas, PyTorch, and advanced deep learning topics such as CNNs, embeddings, RNNs, attention, and transformers.
- Kaggle, owned by Google, is the world's largest data science and machine learning community, offering a rich ecosystem of datasets, Kaggle Notebooks with free GPUs and TPUs, ML competitions with prizes, and courses covering Python, machine learning, and deep learning. Millions of data scientists and ML engineers use Kaggle to explore data, build models, share work, collaborate on projects, and take part in challenges that push the boundaries of AI research and real-world applications.
- An article sharing sources and methods for keeping up with Machine Learning and Deep Learning. The author recommends reputable blogs such as Machine Learning cơ bản, Distill.pub, and Lilian Weng's blog; newsletters such as DataMachina, Papers with Code, and AI Weekly; social networks such as Twitter, Medium, Reddit, and Quora; YouTube channels such as Two Minute Papers, 3Blue1Brown, and DeepMind; and many other quality sources.
- Google Teachable Machine is an interactive in-browser tool that lets anyone train ML models without writing code, using images, sound, or gestures. Models can be exported for web apps and physical devices.
- The release page of d2l-vn (the Vietnamese translation of Dive into Deep Learning) on GitHub, providing the full translation of the D2L book.
- The GitHub repository deepai-book by phamdinhkhanh, collecting deep learning knowledge from basic to advanced.
- A Google Colab notebook for Image Classification practice, a hands-on guide to classifying images with deep learning.
- The GitHub repository with the Python implementation of the Predictive Power Score (PPS) by 8080 Labs, an asymmetric metric that detects linear and non-linear relationships between two columns using a decision tree. Scores range from 0 to 1, a stronger alternative to the correlation matrix for exploratory data analysis.
- A Towards Data Science article introducing the Predictive Power Score (PPS) as a superior alternative to traditional correlation for detecting relationships between variables. Explains how PPS overcomes the limits of correlation by detecting non-linear patterns.
- The scikit-learn documentation page for LabelEncoder, a tool that encodes class labels as numeric values for preprocessing in machine learning.
- A Google search results page for Otsu thresholding, a popular automatic image thresholding algorithm in image processing. Otsu's method computes the optimal threshold by maximizing the between-class variance of foreground and background pixels, and is widely used in medical image segmentation, text recognition, and image preprocessing for computer vision.
- The E-Commerce Data dataset on Kaggle, containing real e-commerce transaction data for data analysis and machine learning projects.
- A Kipalog article on using coroutines in Python to implement a request-dispatching algorithm. The author introduces the concept of coroutines, the difference between synchronous and asynchronous programming, and how to use async/await in Python to handle many requests concurrently without blocking, including a practical example using the asyncio library to optimize performance when sending many HTTP requests at once.
- FastAPI is a modern, high-performance Python web framework for building APIs with automatic interactive documentation (Swagger UI and ReDoc), async support, type validation, and dependency injection, based on standard Python type hints.
- The Python documentation for asyncio Task, a guide to using coroutines and tasks for asynchronous programming in Python.
- An article on Ông Xuân Hồng's blog about XGBoost, the algorithm that won many Kaggle competitions, explaining its principles and applications.
- A ListenData article explaining Weight of Evidence (WOE) and Information Value (IV), important feature transformation techniques in credit scoring and risk analysis.
- The CRBAP agile board on Jira for the Cerebro Lab project, showing tasks and sprint management.
- A database diagram shared on dbdiagram.io, an online tool for designing and visualizing database schemas using a simple DSL code syntax, with collaboration support and export to SQL or PDF.
- A YouTube video tutorial on the EAV (Entity-Attribute-Value) Model in Magento, explaining EAV Entity, Load, and Save.
- An Adobe XD prototype link shared for collaboration and UI/UX design review, letting stakeholders view and interact with the interface design, give feedback, and check specifications in the browser.
- A Viblo article on the Viola-Jones face detection method integrated in OpenCV. The author explains three main components: Haar-Like features for extracting information from images, the Integral Image technique for quickly computing pixel sums over a region, and the AdaBoost algorithm that combines weak classifiers into a strong classifier in a cascade model, reaching roughly 98% accuracy.
- A lecture on the normal distribution on Hoc247, explaining the concept, properties, and applications of the normal distribution in probability and statistics.
- A Vietnam Lab Blog article on types of probability distributions and their properties, including the binomial, Poisson, normal, and more.
- A scientific paper on ScienceDirect about AdaBoost for Feature Selection, Classification, and its relationship with SVM.
- A question on Cross Validated (Stats StackExchange) about using AdaBoost for feature selection, discussing methods and implementation.
- An article explaining in detail the HOG (Histogram of Oriented Gradients) feature description method, a popular feature descriptor in computer vision. The author presents five steps to build a HOG vector: image preprocessing, computing gradients in the Ox and Oy directions, computing a feature vector for each cell by voting gradients into a 9-bin histogram, block normalization to increase invariance to lighting, and concatenating the vectors into the final feature vector.
- A web tool for generating publication-quality neural network architecture diagrams as SVG. Supports FCNN, LeNet, and AlexNet visualization styles with customizable parameters for node size, edge width/opacity/color proportional to weight, and horizontal/vertical orientation.
- A tutorial on visualizing embedding data with the t-SNE algorithm through the Tensorflow Projector. The author walks through the full process: training a CNN on MNIST, extracting 4096-dimensional features from a fully connected layer, reducing to 128 dimensions with PCA, configuring the Tensorflow Projector config file including tensor, metadata, and sprite image, and finally deploying to Github Pages to visualize the results.
- TensorFlow Playground is an interactive in-browser neural network tool from the TensorFlow team, for visually experimenting with neural networks on various 2D datasets. Adjust hidden layers, neurons, learning rate, activation function, and other hyperparameters.
- The Deep Learning course on Udacity (ud730), covering neural networks, convolutional networks, and deep learning fundamentals.
- The free TensorFlow course on Udacity (ud187), a guide from basic to advanced on the TensorFlow framework.
- The Advanced Machine Learning course on Coursera, going deep on advanced ML techniques and real-world applications.
- The GitHub repository learn-machine-learning-in-two-months by bangoc123, collecting the knowledge needed to learn ML in 2 months.
- The Mathematical Foundations for Artificial Intelligence course on Intelligence.edu.vn, providing the math needed for AI.
- The AI For Everyone course on Coursera by Andrew Ng, an introduction to AI for non-technical people.
- Stanford CS20SI: TensorFlow for Deep Learning Research, a course on using TensorFlow for deep learning research.
- An article on nguyenvanhieu.vn guiding beginners through a basic TensorFlow course.
- An article on nguyenvanhieu.vn collecting basic machine learning materials for beginners.
- A Kipalog article sharing the author's journey of teaching themself Deep Learning. It covers a learning path from mathematical foundations such as linear algebra, probability and statistics, and calculus to the core concepts of artificial neural networks, convolutional networks (CNN), and recurrent networks (RNN). The author recommends useful resources such as Andrew Ng's Coursera courses, the Deep Learning Book, and hands-on experience through personal projects.
- Stanford CS231n: Deep Learning for Computer Vision is a well-known course covering convolutional neural networks, image classification, object detection, and segmentation. It has assignments on training neural networks from scratch, lecture videos, and a final project.
- Stanford CS224d: Deep Learning for Natural Language Processing is a course covering word vectors, RNNs, LSTMs, recursive neural networks, and CNNs for NLP. Taught by Richard Socher, now replaced by CS224n.
- The GitHub repository TensorFlow-Examples by aymericdamien, a collection of TensorFlow tutorials and code examples.
- The GitHub repository 100-Days-Of-ML-Code by Avik-Jain, a 100-day Machine Learning learning path with daily code examples.
- A thread on the Machine Learning cơ bản forum collecting ML materials for beginners.
- The Convolutional Neural Networks in TensorFlow course on Coursera, a guide to building CNNs for computer vision with TensorFlow.
- The Convolutional Neural Networks course on Coursera by Andrew Ng, covering CNN architecture, object detection, and visual recognition.
- The Deep Learning Specialization on Coursera by Andrew Ng, covering neural networks, CNNs, RNNs, and structuring ML projects.
- A YouTube video of Lecture 1 of Stanford CS224U: Natural Language Understanding, introducing the NLU course.
- An article on huyenchip.com collecting a free online machine learning curriculum, a high-quality free ML learning path.
- The Deep Learning course on Udacity (ud730), a lesson on neural network concepts, covering the basics of artificial neural networks, backpropagation, gradient descent, and training techniques. The course provides a solid foundation in deep learning fundamentals for anyone who wants to understand how neural networks work and how to optimize training.
- Introduction to Statistical Learning (ISL), a textbook on statistical learning and machine learning fundamentals.
- The Roadmap.sh Engineering Manager Roadmap is a comprehensive step-by-step guide to becoming an Engineering Manager, covering all the skills, knowledge, best practices, and career milestones needed for a technical leadership role.
- A YouTube video guide to a 10-minute at-home chest and arms workout with no equipment, by SHINPHAMM.
- A YouTube video guide to a simple 10-minute at-home abs workout, by SHINPHAMM.
- A VOZ Forums thread with over 2,000 comments on the items with the best price/performance ratio that members have ever bought, including electronics, home appliances, and accessories with real-world reviews.
- A VOZ Forums thread where members share their own professional secrets, page 76, discussing work experience.
- A VOZ Forums thread on skin problems answered by dermatologists, including acne, scars, dark marks, and treatments.
- A VOZ Forums thread on which teabags taste good, where members share reviews and tips for choosing tea.
- A VOZ Forums thread about tea appreciation, where members share experiences and preferences about different kinds of tea.
See also: Learning Resources · LLM Learning Resources · Hardware Rent