Inspiration
When we wanted to learn cybersecurity, every resource felt the same courses, videos, walkthroughs. All theory, no teeth. Real security skills aren't built by reading; they're built by doing. We met at ISRO's Cybersecurity Division, where one of us was building AI to detect network anomalies and the other was designing training programs for government cyber cells and we both hit the same wall. No tool existed that put people in a real fight and let them learn from it.
What it does
Cyber Range pairs a live attacker and a live defender, spins up a virtual machine accessible directly in the browser, and lets them go at each other in real time. No guided steps. No hints. Every move is captured as training data for our AI layer, building a dataset of authentic human decision-making under adversarial pressure that no synthetic source can replicate.
How we built it
We built a real-time matchmaking and VM orchestration system with in-browser terminal access. All session activity is logged and piped into our AI training pipeline. The stack spans full-stack web, cloud infrastructure, and ML systems built across three engineers with backgrounds in AI research, cybersecurity operations, and systems engineering.
Challenges we ran into
Orchestrating live VMs per match at low latency is hard. Keeping sessions isolated, stable, and recoverable while simultaneously capturing granular activity data without impacting performance pushed us technically. Designing open scenarios that are challenging without being chaotic also took significant iteration.
Accomplishments that we're proud of
A working prototype where two people can actually fight each other on a live VM through a browser and every action feeds our AI. No smoke, no mirrors.
What we learned
The best training data doesn't come from labs or simulations. It comes from humans under real pressure making real decisions. That insight shapes everything we build.
What's next for Cyber Range
Scenario-specific environments modeled on real company infrastructure, role-based matchmaking by skill and specialty, and an AI agent trained on millions of live matches, one that can eventually hold its own against human players.
Log in or sign up for Devpost to join the conversation.