What is a good AI project?

AI Projects
,
1

Hello everyone

I am currently in my final year of college and struggling to find a job. I revised my CV and thought about a new project this time. The problem is that I suddenly asked myself, “What is a good AI project?” and realized that I didn’t know the answer. I have learned many models and techniques and practiced with them, but can I say the result I have is a project? Because I do everything very systematically: download online data, process, and train models. I may spend a lot of time to get good accuracy, but all I have is a Jupiter notebook file.

So I want to hear your opinion on what a good AI project is, at least at my level (intern or fresher), and what I should do now. Should I try to do a real app or web with AI integration?

Thank you for reading. I appreciate your answer.

2

I suggest there are two very different types of ‘good’ AI projects. The first, which is primarily what we learn about in these types of courses and specializations, is technical. Discovering a new approach or technique that overcomes some limitation in the state of the art. Success is measured in transactions per second, or number of tokens, map, etc. The second is business oriented, applying existing research and capabilities to solve real world challenges. Success is measured in reduced manufacturing or medical errors, higher crop yields, fewer autonomous vehicle crashes, lower customer churn/higher retention, etc. Both are interesting but to different audiences.

3

A good AI project, especially for an intern or fresher, often involves more than just training models and achieving high accuracy. It should demonstrate your ability to solve real-world problems, apply AI techniques effectively, and communicate your results clearly. Here’s a guide to help you refine your AI project:

Characteristics of a Good AI Project:

  1. Real-World Relevance:
  • Problem Solving: Choose a problem that has practical applications or relevance to industries you’re interested in.
  • Impact: Consider how your project could make a difference or provide value to users.
  1. End-to-End Implementation:
  • Data Collection: Show your ability to gather and preprocess data effectively.
  • Model Development: Demonstrate your skills in building, tuning, and evaluating models.
  • Deployment: If possible, include how you would deploy your model or integrate it into a usable application.
  1. Integration and Application:
  • Web/App Integration: Developing a web or mobile application that uses your AI model can showcase your ability to create a complete product. This might include a recommendation system, chatbot, or image classification tool embedded in an app.
  1. Presentation and Documentation:
  • Clear Communication: Document your project thoroughly, including the problem statement, methodology, results, and code. A well-written report or blog post can highlight your project’s impact.
  • Portfolio: Include your project in a professional portfolio or GitHub repository, with a clear explanation and visualizations of your results.

Suggested Steps:

  1. Identify a Project Idea:
  • Look for problems in domains that interest you, such as healthcare, finance, or social media. Examples include sentiment analysis of social media posts, predictive analytics for stock prices, or a personalized recommendation system.
  1. Develop a Complete Solution:
  • Build a Web/App: Create a simple web or mobile application that integrates your AI model. This could be a chatbot that provides recommendations based on user input or a web app that visualizes your model’s predictions.
  1. Showcase Your Project:
  • Portfolio: Add your project to your personal portfolio or GitHub with detailed documentation.
  • Resume: Highlight the project in your CV, focusing on the problem solved, technologies used, and the impact of your solution.
  1. Gather Feedback:
  • Share your project with peers, mentors, or online communities to get constructive feedback and improve it further.

By taking these steps, you’ll not only demonstrate your technical skills but also your ability to create and present a complete AI solution, which can make you stand out to potential employers.

4

A hybrid of local assistent who you can talk to (whisper model for speech2text for example) and it replies using a text2speech model. Additionally make it control your desktop by voice, kind of Robotic Process Automation, also known as RPA.
RPA is highly demanded, so you will have benefits implementing it “from scratch” to find a good job too.

Another one is Deepl clone running completely locally, with global hotkey, background thread to not block ui, dynamic model selection, UI for settings and so on :slight_smile:

I assume you know programming in general.