SHREYA RAMRAIKA

KIIT, Bhubaneshwar

Bachelor of Technology · Computer Science and Engineering

N/A – June 30, 2026

Academic Global School

Intermediate Examination

N/A – Present

H.P Children's Academy

Secondary School Examination

N/A – Present

Edge-Guided Thermal Image Segmentation using Deep Learning

February 1, 2026 – February 28, 2026

Worked on thermal image segmentation using the SODA dataset, focusing on accurate object boundary detection. Performed data preprocessing and mask alignment to resolve inconsistencies between images and labels. Developed a deep learning-based segmentation pipeline with emphasis on boundary-level prediction. Incorporated edge-guided features to enhance segmentation performance in low-contrast thermal images. Applied data augmentation, normalization, and class balancing techniques to improve model generalization. Trained and optimized the model using PyTorch, iteratively debugging data and model outputs. Evaluated performance using IoU, Dice Score, and pixel accuracy for consistent segmentation results. Developed backend using Flask/Streamlit and created APIs for real-time image processing.

Explainable AI System for Thyroid Diagnosis using Deep Learning and Machine Learning

January 1, 2026 – February 28, 2026

Performed Exploratory Data Analysis (EDA) to identify feature patterns, correlations, and class imbalance. Implemented data preprocessing pipeline (imputation, outlier handling, encoding, normalization). Developed a multimodal AI system combining ultrasound images and clinical data. Fine-tuned ResNet50 (Transfer Learning) for image classification and trained Random Forest on lab features (TSH, T3, T4). Applied feature engineering and late fusion to improve prediction performance. Integrated Grad-CAM for interpretability and visual explanations. Evaluated using Accuracy, Precision, Recall, F1-score, ROC-AUC with threshold tuning. Built and deployed an LLM-based system with backend integration and real-time inference using Streamlit.

LLM-Driven Multi-Agent Research and Report Generation Framework

December 1, 2025 – January 31, 2026

Designed a multi-agent AI system for automated research and report generation. Implemented Retrieval-Augmented Generation (RAG) using FAISS vector database and Sentence-Transformer embeddings for semantic search. Built a hybrid retrieval pipeline combining PDF-based vector search and live web search using APIs. Integrated LLM (Llama 3.1 via Groq) for high-speed inference and response generation. Developed agent orchestration (Planner, Researcher, Writer, Citation) for structured task execution. Applied prompt engineering and reasoning strategies to improve response quality. Built an interactive Streamlit interface with document upload, chatbot interaction, and report generation. Enabled Explainable AI (XAI) by displaying intermediate agent reasoning outputs.

E-commerce Orders Dispatch Automation

August 1, 2025 – October 31, 2025

Analyzed the order dispatch workflow to identify repetitive tasks and automation opportunities. Developed a UiPath-based automation pipeline to streamline order processing for a retail system. Automated data extraction from Excel to identify scheduled and pending orders. Implemented real-time tracking and notification system to update customers on dispatch status. Integrated timestamp-based status updates for improved monitoring and transparency. Generated automated summary reports for warehouse management to enhance operational efficiency.

Python

Udemy

January 1, 2025 – Present

Machine Learning A-Z

Udemy

January 1, 2025 – Present

Automation Testing- RPA

UiPath

January 1, 2025 – Present

Introduction to C++

Coding Ninjas

January 1, 2024 – Present