Flow Anomaly Detection at the Large Hadron Collider using Normalizing Flows

• Applied knowledge distillation to compress a normalizing flow model for real-time inference in L1 Triggers at the LHC to find new physics signals, achieving the inference latency of 10ms required for deployment on FPGAs.
• Submitted to Springer Nature IJSA; awarded Outstanding (O) grade by the University of Delhi.

• Built a CNN-based facial recognition system (TensorFlow / Keras) for automated building access control, covering training through deployment.

• Developed a VGG-19 based CNN model for early COVID-19 detection from chest X-ray images using TensorFlow, demonstrating statistically significant accuracy improvements over traditional diagnostic models.