Projects

Self Projects

AI Support Assistant (Samsung S25 Ultra Support)

Built a Streamlit-based chatbot integrating RAG with PDF manuals.

• Tech: Qwen3-4B-Instruct, QLoRA, RAG, Streamlit
• Fine-tuned Qwen3-4B-Instruct using QLoRA to adopt a patient and professional support persona.
• Delivered real-time, factual assistance with reduced hallucinations and improved user satisfaction by integrating RAG with PDF manuals.

Kinector: A Text-Conditioned 2D Gesture Generator

• Tech: PyTorch, MediaPipe, GRU
• Generated 2D stick-figure animations from text using a <100-sample short-video dataset.
• Built a pipeline with MediaPipe keypoint extraction, GRU pose prediction, and text-to-motion animation.

Course Projects (Computer Vision & GenAI)

Medical Image Deblurring

• Tech: Scale-recurrent network, Spatial-asymmetric attention
• Mitigated motion blur in multi-modal medical images (30% of scans affected) to improve diagnostics.
• Improved clarity by 24% (PSNR), providing higher-quality inputs for reliable medical analysis.

Learning with Noisy Labels using Vision Transformer (ViT)

• Classified images with 40% label noise on CIFAR-100 by applying the state-of-the-art Turtle method.
• Achieved 83% accuracy with a Vision Transformer, maintaining strong performance despite noise.

Fine-Grained Classification on CUB Dataset

• Designed a CNN with <10M parameters, achieving 35% lower model size with competitive accuracy.
• Attained 87.14% top-1 accuracy on 200 bird species.

Autoencoding Beyond Pixels

• Implemented VAE/GAN to generate high-fidelity images, improving perceptual score by 18% over SOTA.
• Achieved superior fidelity via latent space arithmetic on 10K+ images.

Achievements

• GATE 2022: Attained 97.84 percentile in Engineering Sciences (XE) (Top 15k candidates).
• GATE 2022: Attained 96 percentile in Mechanical Engineering (ME) (Top 80k candidates).