Inspiration
As automotive enthusiasts always pursuing new knowledge of our favorite brand or model, sometimes obtaining obscure information can be very difficult. By creating a central hub of automotive data, we can get the information we need as enthusiasts need to make informed decisions or learn something new.
What it does
By querying for information in an everyday language such as: "get me some information on a Porsche 911", we pull out relevant information on the specific model and make in question from over 40,000 vehicles ranging from the model year 1984 to current.
How we built it
We used React.js for our frontend and AWS services for our backend. React.js allowed us to create a clean landing page for our project and AWS Amplify gave allowed us to call our custom API from the frontend. Our backend architect consists of an API gateway to facilitate traffic to our lambda functions. AWS Lamda was used to drive all backend logic. We connected Wit.ai to lambda using wit's python library. All natural language processing was done on Wit.ai. Once lambda had all the necessary information, it would send it back to our frontend and React.js would display it.
Challenges we ran into
While developing Drivr.space, we ran into many beginner errors having to learn new frameworks and hosting with AWS Lambda, and challenges such as finding large quantities of vehicle information formatted in a way easily and predictably navigated. For example, due to differences in the file systems of AWS Lambda and machines which we developed the back end of our website, we had to utilize different means of loading and decoding a file which our program could interpret.
Accomplishments that we are proud of
We are very proud to have come into BorderHacks not knowing ReactJS, or AWS lambda and through hard work and dedication to creating something we are proud of. Additionally, we are proud of learning something new, and teaching each other.
What we learned
We have learned in the process of developing Drivr.Space useful information such as utilizing natural language processing, ReactJS, AWS Lambda, and the skill of recognizing a problem and working towards a solution methodically, while addressing potential issues away to create a more robust program.
What's next for Drivr.Space
We would like to train the API we use to do natural language processing to drive Drivr.space to handle additional questions and reply more fluidly with specifics. We would also like to see the order complexity of the backend of Drivr.Space to decrease. Because our data set is so broad and our data structure is inefficient, every call requires over 5 million operations, making the program very slow. We see this problem being remedied by storing our data in a more efficient data structure for our application, such as a relational database.
Log in or sign up for Devpost to join the conversation.