Gradual | Devpost

Landing Page
Dashboard

Inspiration

As a team of students from the University of Pennsylvania, we find that in the midst of all our activities, clubs, psets, social events, it's often very hard to keep a clear vision of our accomplishments and goals while balancing a healthy lifestyle. This sparked our interest in creating Gradual, an assistive tool that takes up the hefty task of personal organization.

What it does

Auto-capture meaningful moments from digital behavior like messages, meetings, and purchases. No need for manual journaling or tagging ever again.
Extract and synthesize important patterns from various buckets of user data, which are turned into individual nodes related to each other on a timeline — revealing connections you didn’t know existed.
Interactive user interface with dynamic timeline navigation and visual cues to highlight day-to-day developments.
Leverage AI to uncover hidden insights in user data. Rediscover your habits, interests and breakthroughs like never before.
Surface emotional rhythms by detecting patterns in your productivity, communication, and engagement levels across time.
Identify points of improvement based on well-structured time data of previous events and happenings.
Provide data-backed suggestions of future goals for clearer career and personal planning.

Challenges we ran into

First hackathon experience for a member
Leveraging Google Cloud API to parse user data into usable chunks of data for analysis.
Learning how to use new libraries such as React Flow to create our graph-based timeline interface
Integrating Modal to adopt our custom model

Accomplishments that we're proud of

Integration of diverse features such as graph organisation, multimodal data synthesis and AI into a cohesive organisational tool
Creating a reactive timeline system that automatically reorganises itself and reconnects nodes based on user input
Designing an easy-to-use navigation system to explore the complex timeline

What we learned

How to transform unstructured, messy user data into meaningful, structured insights through clustering, embeddings, and time-based filtering.
How to connect multiple APIs and frameworks cohesively — from parsing raw data with Google Cloud to rendering interactions with React Flow.
How to balance functionality and performance when working with large-scale data across real-time updates and visual re-rendering.

What's next for Gradual

Even wider variety of data sources that allow for precise tracking.
Simulations of possible future timelines centered around user-inputted goals and accomplishments
Extending beyond the student demographic, with custom data extraction models geared towards group organisation, educational venues or even vertical integration

Presentation: https://drive.google.com/file/d/1Pe9zCM68RANTZNFf5qALlyszQfhtSJBA/view?usp=sharing

Team members: Ryan Tanenholz, Samuel Lao, Willard Sun, Yuvraj Lakhotia