The world is becoming increasingly interconnected, and with that comes the need for accurate and culturally sensitive translations. We recognized the opportunity to leverage the advancements in machine learning and natural language processing to create a translation app that can understand and adapt to the complexities and nuances of different languages and cultural contexts.
What it does
NativeSay is a translation app that helps users understand the context in which translations should be used, and provides culturally and contextually aware translations based on the context the users specify. Our goal is to facilitate effective communication in different languages and cultural contexts.
How we built it
We built the backend using Django and the frontend using NextJS with Typescript, along with Material UI for the design. We leveraged the GPT-3.5 language model to power our translations.
Challenges we faced
On the frontend, we struggled to understand the intricacies of NextJS and to modify the components in Material UI. Additionally, integrating the frontend with the backend proved to be a challenge. On the backend, we encountered a CORS issue during the final hour of development, which caused some panic. Additionally, due to the probabilistic nature of the model, it was challenging to ensure consistent output format and we had to put in a lot of effort in fine-tuning and prompt engineering to make the model reliable.
Accomplishments we are proud of
Despite the challenges we faced, we are proud of the fact that we were able to develop the main app UI and to query the backend correctly. Furthermore, we are proud that we managed to display the desired output correctly. The translation model is not perfect, but during the demo to the judges, we ran about 30 translations and only 2 or 3 of them were problematic. We are also proud of the fact that we received positive feedback from the judges and many expressed interest in being a user of our product.
What we learned
We learned that patience is a virtue and that coding and sleep don't always go hand in hand. We also learned about the importance of proper planning and time management to overcome challenges during the development process.
What's next for NativeSay
Our next step is to improve the user interface and add authentication features. We received a lot of interest from potential users during the hackathon and we plan to launch the product for beta testing in the very near future. Keep an eye out for updates!