Inspiration
For all the ML developers out there, getting the dataset that matches the criteria for your Machine Learning Model is hard. We get it. That is why we created Auto Bot to help you buy/auction/sell the dataset of your choices.
What it does
AutoBot is the ultimate marketplace for all the ML developers looking for the right kind of resources for your apps.
Find & Search for datasets that can
- Train your model
- Test your model
- Develop your model
Buying & Selling datasets have never been easier, until now!
AutoBot helps you find the resources you need without wading through reams of information on the internet- saving your valuable time as a developer and making you more productive than ever.
How we built it
Web-App- Our web-app is built with React.
Algolia Javascript API Client- A flexible AI-powered Search & Discovery platform
Algolia’s search-as-a-service allows our web-app to easily develop tailored, fast Search & Discovery experiences that bring together developers, researchers, institutions under a single platform.
Indexing- The search feature works just like any other standard search, a keyword as an input is accepted from a search bar and compared against schemaless objects
with a unique objectID
associated with them. These objects are stored as "Records" in our database of datasets.
Stripe Node Library The checkout pages are built with stripe, we pay for the hardwork done in creating the datasets and models
Firebase Cloud Firestore and Authentication
Challenges we ran into
- No prior experience with Datasets and Machine Learning, thus hard time figuring out the way, our pod leader and members came to rescue us.
- Automating demo-dataset discovery.
- Parsing JSON correctly
- Working with Algolia and indexing
- Indexing and updating firebase to make sync with both datastores
- Building good design and we messed up with the frontend at start
- Making our app more secure and less attack vulnerable
- Netlify build issues and page redirects
Accomplishments that we're proud of
- Nightly builds
- Deploy Previews In PR
- Github actions and Github project management
- Lightning-fast payments integration
- Implementing search into our platform with Algolia
- Our Logo
What we learned
- Working and collaborating in team with github, discord and notion
- Learnt more about React and Frontend
- How to design database schema for the features and further modify it to cover the use cases
- Algolia sdk for react and react-instant-search feature
- Firebase realtime db and firestore CRUD operations
- Google OAuth
- Integrating search feature into a platform
- Adding a payments gateway into a platform with stripe
- Dfinity's Internet of Computer platform & the Canister SDK
What's next for AutoBot
AutoBot CLI- A CLI is always lightweight, robust, and faster in comparison to its web-app counterpart. A clean CLI that can let any developer get the right datasets by typing only minimal commands would boost their productivity and efficiency while getting their ML-Apps up and running.
Robust Developer support: We are already providing underlying code support with the datasets to kickstart the building and development in snap, we are about to extend it to benefit every new developer to start learning ML and getting in love with it by our extended code support and exensive theory.
Github like platform for Data engineers: We have a strong motivation to convert our app to build a new github specifically for Machine Learning and DataScience, so that everyone can collaborate, build and grow.
Autobots, roll out!
Log in or sign up for Devpost to join the conversation.