Final year Computer Science student

darshan kumar  • Peerlist

Built a cloud-based real-time monitoring system using Google Earth Engine, Vertex AI, and BigQuery. Deployed anomaly detection algorithms to identify deforestation patterns and generate alerts with sub minute latency.

AI-Enhanced Cloud-Based Forest Monitoring Sys

Developed a graph-based system using Python, NetworkX, and Genetic Algorithms to maximize user influence on Instagram. Incorporated behavioral insights and performance optimization techniques to outperform traditional influence models.

Behavior-Aware Influence Maximization on Inst

A real-time malware detection framework for Industrial Control Systems using hybrid SARIMA-LSTM models. The system detects cyber-attacks targeting sensor data and PLC logic by analyzing process-layer anomalies. SARIMA captures seasonal trends while LSTM learns complex temporal patterns. Validated on SWaT and HAI datasets with 5.5 hours early warning capability. Lightweight 199KB model suitable for edge device deployment in critical infrastructure.

Industrial Control System Malware Detection v

Designed a self-attention-augmented 1D CNN with TensorFlow and Keras for Android malware detection. Achieved 92% accuracy by modeling long-range opcode dependencies and proposed explainability for model predictions

Interpretable Malware Detection Using Self-At

Developed a responsive full-stack web application to manage live cricket scores using React.js, TailwindCSS, and REST APIs. Implemented an auto-refreshing system for real-time updates, integrated winning probability prediction models, and designed a user-friendly dashboard for scorekeepers.

Live Cricket Score Management

Developed Parsely AI, a Gemini-powered document processing platform that uses Large Language Models for intelligent claim evaluation and semantic search across unstructured documents like contracts and policies. Built with FastAPI backend, Streamlit frontend, and cost-effective architecture delivering 90% savings over traditional AI solutions

Parsely AI

Built a deep learning-based image classification model using CNNs to detect retinal diseases (cataract, glaucoma, diabetic retinopathy). Improved diagnostic accuracy and supported healthcare automation through medical dataset analysis

Retinal Disease Detection using Deep Learning

A comprehensive Android productivity application designed to consolidate multiple student management tools into a single, unified platform. Student.dev addresses the fragmentation problem of using 5-8 separate apps for notes, attendance, habits, and budgeting by providing an offline-first, feature-rich alternative built with modern Android architecture patterns.