During my AI internship at IRDE, DRDO - Ministry of Defence, Govt. of India, I fine-tuned YOLOv11n on a custom image dataset for FireWatch. I also developed a computer vision pipeline with real-time inference, live analytics dashboards, and automated alerts.
I built 3D-FY, an end-to-end pipeline that converts 2D video into anaglyph, side-by-side VR, and depth-map outputs. I improved its depth estimation and temporal smoothing, and deployed it as an interactive Gradio application.
With CanvasAI, I combined FLUX.1-Fill, FLUX.1-Kontext, and Real-ESRGAN for outpainting, colorization, and super-resolution. My other projects include automated video color grading and video anonymization; a 3D-FY book chapter was published at NCRIETS 2026.
