Inspiration
Kivy was inspired by the realization that language barriers often pose a significant challenge for immigrant workers in the job market. We wanted to find a way to break down these barriers and create a more inclusive work environment in which talented individuals from all backgrounds have an equal opportunity to succeed. We created a tool that can accurately generate reports in any language using NLP models, making it easier for managers and coworkers to understand worker routines and the work done.
What it does
Kivy is a report-generation tool that uses natural language processing (NLP) models to generate accessible report PDFs from various input sources such as text, voice memos, and images in a foreign language. The tool translates any language in real-time, allowing managers and coworkers to better understand worker routines and work. Employees and managers can ask questions about previous reports and get instant answers using the retrieval-augmented generation pattern.
How we built it
Kivy was built with NLP models such as machine translation, image recognition, and speech recognition. To ensure high accuracy in generating reports from various input sources, we trained the models on large datasets. We built the application in Python and the API that connects to the NLP models.
Challenges we ran into
One of the major challenges we encountered was prompt engineering and accounting for the hallucinations. We also had to deal with issues like real-time translation and making sure that the PDF reports generated were easily accessible to everyone. But also that the translations match the output of the languages in a professional manner.
Accomplishments that we're proud of
We are proud to have created a tool that has the potential to significantly improve the lives of immigrant workers. By removing language barriers, we make it easier for them to find better jobs and improve their quality of life. We're also proud of having created an innovative solution that can aid in the creation of more inclusive workplaces where everyone has an equal chance of success.
What we learned
We learned a lot about NLP models, such as machine translation, image recognition, and speech recognition, while developing Kivy. We also learned about the difficulties in developing a tool that requires real-time translation and the generation of accessible PDF reports.
What's next for Kivy - Inclusive workplaces, one language at a time!
Kivy's capabilities will be expanded to other industries and use cases in the future, including healthcare and education. We also intend to improve the accuracy of the NLP models and expand the application's features, such as image and video (OCR) translation. Our ultimate goal is to create a world in which language barriers no longer limit people's opportunities and in which everyone has equal access to employment and education.
Log in or sign up for Devpost to join the conversation.