The problem Tareek Pe Tareek solves

Tareek Pe Tareek addresses the challenge of making meaningful connections by analyzing over 45 user parameters to generate highly accurate match recommendations given by an intelligent system. Parameters were assigned different weights and a final recommendation score is used to display the best Profiles. The platform facilitates real-time communication through WebSockets for instant messaging and includes a unique fantasy page where users can message anonymously and share their thoughts. Additionally, users can easily request dates, enhancing interaction. The site also features a well-trained chatbot that incorporates extensive data, including watch-out information, and is tailored specifically for the IITR Junta. Although the functionality for storing dates as albums with photos and descriptions was not implemented due to time constraints, the platform still provides a comprehensive and engaging solution for modern dating. We have also added a moderator in the fantasy page that automatically puts asterisks for cuss words (cuss word detector). We have also added a Love chatbot. To fine-tune and build a chatbot for our dating website, we leveraged advanced NLP techniques to create engaging conversations that can help users go along.

Challenges we ran into

We could not implement all of our ideas, despite having knowledge ; some of them were :

1. A two player game built using Django WebSockets to play across, & encourage bonding between potential partners.
2. Wanted to create personalized Date history (we even coded the models and the serializers but because of time issues we could not put it to the frontend)
3. We made the chatbot but could not integrate it with our website due to the time constraints.
4. In the chatbot making, we could use infer API key of the popular LLM's that are available publicly, but we didn't ; tried a finetune a big number of parameters of existing small LLM's to change it according to our usecase.
5. The model for chatbot was over 5GB in size, we couldn't upload it on our github repo.