RagulRajkumar User

Built an end-to-end Retrieval-Augmented Generation pipeline that answers questions grounded in uploaded PDF documents using FAISS retrieval and a Qwen2.5-1.5B-Instruct LLM, with chunking and context injection to reduce hallucinations. Deployed an interactive Gradio app on Hugging Face Spaces for side-by-side evaluation with and without retrieval context.

Developed a real-time object detection REST API using YOLOv8 (Ultralytics) with FastAPI, producing bounding boxes and confidence scores for 80 COCO classes. Added JSON detection and annotated-image outputs and integrated a Gradio interface; migrated deployments from Render to Hugging Face Spaces for stable inference.

Fine-tuned Stable Diffusion v1.5 using LoRA on the Hugging Face anime-faces dataset (5,000 images), reducing trainable parameters from 860M to 797K (0.09% of model weights). Tuned LoRA rank/alpha on targeted attention layers and benchmarked base vs. fine-tuned outputs to document domain adaptation results and limitations.

Created an interactive OpenCV-based image-processing tool in Colab with ipywidgets, implementing 13 filters spanning edge-detection, artistic effects, color-tone, and portrait categories. Built upload-to-filter-to-download UI and implemented custom convolution kernels and pixel-level transforms (e.g., Gaussian blur, sharpening, embossing, sepia, and GrabCut background blur).

Performed data validation and automation using advanced Excel to extract actionable business insights. Built an interactive Tableau dashboard with KPI visualizations while gaining exposure to professional data reporting standards and forensic technology workflows.

Pursuing an M.E. in Computer Science and Engineering with a minor in Big Data Analytics. Focuses on applied AI and data science through end-to-end ML/DL projects including time series forecasting, diffusion models, and computer vision.

Grade: CGPA: 8.36

Completed a B.E. in Computer Science and Engineering with a CGPA of 8.36. Built foundational knowledge in computing and software development to support later ML and data science work.