28 production‑ready AI projects on GitHub

A GitHub repository called AI-Project-Gallery is getting attention for a simple reason: it does not pretend that an AI project ends when a model trains. The collection, maintained by Hema Kalyan Murapaka under the handle KalyanM45, gathers public projects that run from input to output. GitHub’s current listing shows 31 named projects in the repository, spanning machine learning, generative AI, computer vision, recommendation systems, web scraping, Power BI, and healthcare-flavored classification work. Several are explicitly marked “End-to-End,” which is the point of the whole exercise. (github.com) That distinction matters because most people still learn AI through fragments. They get a notebook. They get a dataset. They get a model that reaches an accuracy number. Then the lesson stops. This repository is useful because it pushes past that stage into the messier parts that decide whether anything can actually ship. The project list includes things like Airbnb price prediction, chest disease detection, movie recommendation, spam email detection, a market insight workflow, and a command-line documentation tool called Doclify. Those are not all equally sophisticated, but they do force a builder to think about data flow, interfaces, dependencies, and what a finished application looks like when someone else has to run it. (github.com) The repository’s appeal is also in its mix. Older AI learning paths often split the world into narrow tracks: classical ML over here, computer vision over there, LLM apps somewhere else. AI-Project-Gallery jams them together. A reader can move from regression and classification to Gemini and OpenAI chatbots, then into healthcare assistants and agentic workflows, without leaving the same codebase. That makes the collection less like a course and more like a working portfolio. GitHub’s topic pages now surface it prominently under both computer-vision-projects and generative-ai-projects, which helps explain why it has spread so quickly. (github.com) The numbers show that spread. GitHub’s latest cached page lists the repository at about 4.6 thousand stars and more than 900 forks, with 51 commits and a single public contributor. The maintainer’s profile describes him as a generative AI engineer and product associate based in Andhra Pradesh, India, and points to other related repositories, including MarketInsight, Doclify, and OpenHealth. In other words, this is not a giant foundation-backed open-source framework. It is one person turning portfolio work into a public map for other engineers. That is part of why it resonates. It feels attainable. (github.com) What the repository does not do is magically certify anything as “production-ready” in the strict enterprise sense. GitHub’s cached README is a catalog, not an audit. It shows project names, domains, repository links, and which entries are labeled end-to-end, but it does not prove that every app has hardened security, monitoring, CI/CD, scaling tests, or long-term maintenance. The useful claim here is narrower and more believable: these are fuller application references than the average isolated notebook, and that alone can shorten the distance between tutorial knowledge and deployable systems. (github.com) That is also why the “coming soon” section may be the most revealing part of the page. Beneath the current list, the maintainer names 10 more planned projects, including deep fake detection, driver drowsiness detection, arrhythmia disease detection, diet recommendation, breast cancer detection, kidney disease detection, text summarisation, brain tumor detection, pneumonia detection, and realtime face detection. The list reads like a roadmap from classroom demos toward harder, more consequential systems. The concrete detail is almost plain enough to miss: “Deep Fake Detection” sits at number one. (github.com)