Code Genie

CodeGenie Logo
LogIn
ProblemDefinition
ProposedProblem
UserProfile
AboutUs
Contact

Inspiration

Many programming learning and practice platforms offer generic exercises that do not adapt to users' specific skills, goals, and preferences, while programmers are looking for high-quality resources to learn and achieve their knowledge objectives.

What it does

We build a user's profile by getting inputs on examples of code they made or even their CV in order to identify their profficiency in programming languages and indentify key concepts they already know, this so we can offer practice problems that represent a real challenge for the user and identify what they want to learn and fulfill their needs.

How we built it

The interface was done using NextJS, MongoDB was used for the database, Javascript to get abstracts from the PDFs and Python to use and test SofttekLLMSDK.

Challenges we ran into

We faced difficulties getting our LLM models to work how we wanted them to behave. After several iterations, we found the optimal prompt to get results that were consistent enough to prove our concept.

Accomplishments that we're proud of

We´re proud of retrieving information (text) from a PDF in order to use code samples in SofttekLLMSDK to analyze and check user´s profficiency in different programming languages. We´re also proud of getting the optimal prompt to get consistent results for code grading and problem generating.

What we learned

We learned how to use SofttekLLMSDK which allowed us to not only test LLM models but also learn about how they work and how every parameter affects the behaviour of the model. We also learned how to extract text from PDF files using javascript.