Sync | Devpost

GIF
GIF
Landing Page

Inspiration

T-Mobile needed a way to see network health and customer sentiment in one place. NVIDIA challenged us to go beyond chatbots and build reasoning agents that plan and act. We combined both goals to create an AI system that senses problems before users complain and repairs itself through simulation.

What it does

Sync AI monitors live and external data streams like customer reviews, network metrics, and API sources. It detects anomalies in sentiment or performance, predicts issues, and simulates the best repair plan before human intervention. The dashboard displays real-time insights, including customer churn, regional performance, and sentiment trends.

BEST Design

Sync features a modern dark-themed interface with pink and purple accents that reduce eye strain during extended monitoring. The dashboard utilizes real-time visualizations, including interactive heatmaps, trend graphs, and color-coded status indicators, to make complex network data immediately understandable. The layout prioritizes critical information with smooth animations that highlight important changes. The responsive design ensures seamless monitoring across all devices, transforming network operations into an intuitive, visually engaging experience.

How we built it

We used NVIDIA Nemotron for agent reasoning, Brev for GPU container deployment, and multiple APIs for live data integration. Multi-agent workflows handle different tasks: one analyzes sentiment, one detects technical issues, one simulates fixes, and a governing agent prioritizes actions. The frontend uses a live dashboard for visualization with heatmaps, graphs, and customer metrics.

Challenges we ran into

The hardest part was synchronizing multi-agent behavior and aligning simulation timing with real-time data flow. Integrating diverse APIs into a single adaptive system required careful coordination. Getting meaningful sentiment data from noisy sources also took multiple refinement loops.

Accomplishments that we're proud of

We built an agentic system that not only analyzes but also acts. The simulation-driven self-healing network concept is a breakthrough in network intelligence. Our dashboard creates full visibility into both customer and system experience simultaneously.

What we learned

We learned how to structure agent reasoning loops using Nemotron’s ReAct pattern and manage real-time data pipelines efficiently. Building a simulation that mirrors real-world decision tradeoffs taught us how to align AI reasoning with tangible business impact.

What’s next for Sync AI

We plan to scale Sync into a real-time T-Mobile operations layer that automates network repair, customer experience insights, and predictive maintenance. Future updates will connect live tower telemetry, deploy autonomous patching, and evolve into a full adaptive intelligence ecosystem.