ABHISHEK KUMAR SHUKLA

Analyzed denoising and built transformer-based cybersecurity anomaly detection models using custom Darknet and malicious script datasets with scalable data pipelines and feature engineering. Fine-tuned PyTorch/HuggingFace models to improve zero-day threat detection performance through hyperparameter tuning and benchmarking.

Developed production-scale pipeline for GSDIFF research in floorplan generation. Hand-coded edge transformer, eliminated unnecessary edges and improved pipeline efficiency using K-means clustering.