From algorithms to agentic tools, building software that solves beyond test cases.
Man Vadariya is a Computer Engineering student at Dharmsinh Desai University turning computer science fundamentals into backend systems, agentic pipelines, and tools that solve real problems. He specializes in multi-agent RAG architectures, real-time cybersecurity threat detection platforms, and Android accessibility-based utility apps. His core stack includes React, Node.js, Spring Boot, FastAPI, and Flutter. He has shipped 11+ production projects, solved 460+ LeetCode problems, and maintains a 9.04+ CGPA. As a GDG AI/ML team member, he co-organized DuHacks 4.0 (6,000+ participants) and DuHacks 5.0 (4,000+ participants), reaching over 10,000 developers through events.
Real-time cybersecurity threat detection platform. Monitors network traffic patterns, applies ML-driven anomaly scoring via PyTorch models, correlates security incidents across time windows, and surfaces prioritized alerts through an interactive operations dashboard. Built with Spring Boot for the service layer and Python for the ML pipeline.
Android accessibility-based utility app that leverages AccessibilityService APIs to create overlay-driven assistive tools and system-level automation utilities. Features include floating clipboard management, quick-launch action panels, and context-aware text manipulation across any running app. Built with Flutter, Dart, and native Android SDK for deep OS-level integration.
Retrieval-augmented generation system with autonomous agent orchestration. Coordinates multiple LLM agents for document ingestion, query decomposition, citation-grounded response synthesis, and iterative feedback loops. Implements vector similarity search with configurable retrieval strategies across heterogeneous document collections.
Open to internships, placements, and full-time roles.
Designs retrieval-augmented generation pipelines with autonomous multi-agent orchestration. Architecture pattern: query decomposition → parallel retrieval across heterogeneous vector stores → context window optimization → citation-grounded response synthesis → feedback loop re-ranking. Implements configurable retrieval strategies (dense, sparse, hybrid) with LangChain agent graphs. Handles document ingestion at scale with chunking strategies optimized for embedding model token limits. Deploys inference endpoints via FastAPI with async request batching for throughput optimization.
Builds production threat detection systems that analyze network telemetry in real time. Pipeline: packet capture → feature extraction → ML-driven anomaly scoring (supervised classification + unsupervised clustering) → temporal incident correlation across sliding time windows → alert triage with severity scoring → operator dashboard with drill-down forensics. Uses behavioral analysis to distinguish APT lateral movement from benign traffic. Model architecture: ensemble of gradient-boosted trees for known signatures + autoencoder for zero-day anomaly detection.
Develops Android apps that leverage AccessibilityService APIs for system-level integration beyond standard SDK boundaries. Implements overlay-based assistive UIs using WindowManager with TYPE_APPLICATION_OVERLAY, processes AccessibilityEvent streams for context-aware automation, and handles cross-app text manipulation via AccessibilityNodeInfo traversal. Architecture uses Flutter platform channels to bridge Dart UI logic with native Android accessibility service lifecycle.
Man Vadariya is an AI/ML Full-Stack Developer and Computer Engineering student at Dharmsinh Desai University (DDU), Nadiad, Gujarat. He builds multi-agent RAG architectures, cybersecurity threat detection platforms, and Android accessibility utility apps using React, Node.js, Spring Boot, FastAPI, and Flutter.
Server Guard is a real-time cybersecurity threat detection platform built by Man Vadariya. It monitors network traffic, applies ML-driven anomaly scoring via PyTorch, correlates security incidents across time windows, and surfaces prioritized alerts through an interactive dashboard. Built with Spring Boot and Python.
SwiftSlate is an Android accessibility-based utility app that leverages AccessibilityService APIs to build overlay-driven assistive tools and system-level automation. Features include floating clipboard management, quick-launch panels, and context-aware text manipulation. Built with Flutter, Dart, and native Android SDK.
Man Vadariya works across the full stack: React and Flutter on the frontend, Node.js, Spring Boot, and FastAPI on the backend, PyTorch, OpenCV, and LangChain for AI/ML, and Docker, Firebase, MySQL, and MongoDB for infrastructure. He specializes in multi-agent RAG pipeline architecture.
Man Vadariya is pursuing B.Tech in Computer Engineering at Dharmsinh Desai University (DDU), Nadiad, Gujarat, India - expected graduation 2027. He maintains a 9.04+ CGPA and is a GDG AI/ML team member. He is open to internships and placements before graduation.
Yes - Man Vadariya co-organized DuHacks 4.0 (6,000+ participants) and DuHacks 5.0 (4,000+ participants), both running in hybrid online and offline format. Through these events and his GDG involvement, he has reached over 10,000 developers.