Project 1: RAZOR – An Autonomous Threat-Evasion AI for Aerospace Defense

Short Description:
How I’m using Kalman filters, Reinforcement Learning, and simulations to build an AI that can dogfight like a fighter jet pilot. Inspired by DARPA's AlphaDogfight and Jane Street models.

Imagine a jet that doesn’t just fly — it learns, adapts, and survives.

That’s what I’m building with RAZOR, my personal research-grade AI system designed for autonomous threat detection and evasive maneuvering in adversarial aerospace environments.

The Idea:-

Inspired by quantitative finance trading systems (like those used at Jane Street), RAZOR uses real-time optimization and predictive models to detect incoming threats — be it missiles, enemy UAVs, or electronic jammers — and dynamically choose survival strategies. Instead of "buy low, sell high", RAZOR plays “dodge fast, strike smart.”

Key Tech:-

• Kalman Filter for sensor fusion and object tracking

• Bayesian inference for probabilistic decision-making

• Reinforcement Learning to evolve better evasive maneuvers

• Monte Carlo simulations for predicting enemy trajectories

• A future plan to embed it in Unity-based or ROS-based flight simulations

My Vision

This isn’t just a school project. I treat RAZOR like a PhD thesis — long-term, evolving, and serious. By the time I start my Master’s in Aerospace, RAZOR will already have years of groundwork behind it.