The problem MedAssist solves

Medassist is a conversational AI chatbot that assists people in finding medical information quickly and easily. By utilizing natural language processing and machine learning algorithms, Medassist can provide users with personalized responses to their inquiries.

The platform uses OpenAI's GPT-3 language model and Pinecone's similarity search index to generate answers to user queries and provide relevant information. Users can input their questions into the web interface, and the application will generate a list of relevant responses, allowing users to access the information they need quickly and conveniently. The platform is user-friendly and accessible to individuals of all technical abilities, making it an ideal resource for anyone seeking accurate medical advice.

Medassist seeks to bridge the gap between individuals and reliable medical information, providing a comprehensive and accessible solution to the challenges of navigating complex healthcare systems. It aims to provide personalized and interactive healthcare advice to patients by answering their queries based on their symptoms and medical history.

Challenges I ran into

One of the major challenges I ran into while making MedAssist was Fine-tuning the language model to ensure accurate and relevant responses. To overcome this, I referred to OpenAi's Documentation and also experimented with different hyperparameters and fine-tuning strategies for the language model.
Another problem was integrating OpenAi's GPT-3 Model into our project and using Pinecone's similarity search index and integrating the dataset with it. Again, to overcome this, I used Both Pinecone's documentation as well as OpenAi's davinci model's documentation for it.
The next issue came when I tried to deploy MedAssist into vercel, since the vercel free hosting has a filesize limit of 50mb whereas my main.py file exceeded 90mb. Hereby I spent a lot of time optimizing the size of my file by reducing the space complexity of my code.