Devgna Vyas — Systems-First Engineer

Profile

Systems that hold up.

I am a BCA student at Charotar University of Science and Technology (CHARUSAT), Smt. Chandaben Mohanbhai Patel Institute of Computer Applications (CMPICA), specializing in systems-level architecture, backend performance, and cybersecurity protocols.

My approach is first-principles driven: I believe in understanding standard RFC specifications, parsing data streams, evaluating packet flows, and studying systems failure paths before writing code. I utilize AI tools and automated agents as execution multipliers to accelerate design cycles, but keep complete logical ownership of architecture.

Cybersecurity is a serious area of focus for me, especially network intrusion, Linux kernels hardening, and local protocol validation. I build code that holds up under load and prevents vectors of abuse.

Affiliation BCA, Computer Applications
CHARUSAT · Batch of 2026

Location Coordinates Nadiad, Gujarat, India

Primary Fields Backend Engineering · Network Security · Linux Internals
Quantized AI Models · Mobile Systems

Verified Certifications Google Crash Course on Python
Oracle Java Foundations
Tools of the Trade: Linux & SQL (Google)

Case Studies

Projects built with depth.

Featured Mobile Engineering Case Study

Same.Energy Android Client

An elegant, gesture-driven client built using Flutter and Dart for Same.Energy's semantic visual search engine. Replicates custom web-platform aesthetics in a highly-optimized mobile framework.

The Challenge

Translate a complex semantic image query web search into a physics-driven responsive mobile grid, handling asynchronous image pre-fetching and encrypted local caching on strict memory boundaries.

What Was Built

A full client integrating masonry visual grids, curated user feeds, and collections folders. It leverages physics-based scrolling and custom gestures with full-screen viewer overlays.

The Architecture

Decoupled Clean Architecture. Employs Riverpod for predictable reactive state boundaries, Dio client for network request intercepts/timeouts, GoRouter for type-safe navigations, and secure local keychain storage.

Flutter Dart Riverpod GoRouter Dio Client Clean Architecture

Inspect Repository Code ↗

Biometric IoT Systems Study

Smart Attendance System

A secure, centralized, multi-factor attendance verification platform designed for academic institutions to eliminate proxy marking and location spoofing.

The Architecture

Built a high-performance backend using Go and Gin, leveraging PostgreSQL for durable records and Redis for fast transient session state. Facilitated real-time bidirectional communication via Gorilla WebSockets for live attendance dashboard updates.

Tri-Factor Verification

Engineered a robust anti-spoofing pipeline combining rapidly rotating HMAC-SHA256 signed QR codes, on-device Google ML Kit face liveness checks (blink/smile detection), and BLE beacon proximity handshakes via a cross-platform Flutter application.

Security & Cryptography

Implemented RS256 asymmetric JWT token signing, Argon2id password hashing, and mDNS/UDP LAN auto-discovery. Designed an audit-ready architecture with automated PDF/Excel report generation ensuring tamper-evident attendance validation.

Go Flutter PostgreSQL Redis Google ML Kit

Inspect Repository Code ↗

Academic Publication

Research under constraint.

Springer CCIS Series Publication

HERMES: Design and Deployment of a Hybrid AI/ML Network Security System on ARM Clusters for Edge Environments

Arpankumar G. Raval & Devgna Vyas

Presented at 7th International Conference on Soft Computing and Its Engineering Applications (icSoftComp 2025)

📅 Dec 2025 📚 Springer CCIS Volume 🏛️ CHARUSAT Research Lab

Abstract: Designed and deployed a lightweight, cooperative intrusion detection system (IDS) combining signature-based filtering with a quantized 3-layer Deep Neural Network (DNN) classifier. Optimized specifically to run inline on low-power ARM edge nodes, securing local edge topologies without compute bottlenecks.

Evaluation Metric	Benchmarked Outcome
Detection Accuracy	94.7% (Novel attack identification)
Low-Power Envelope	4.2W per ARM active node
Packet Throughput	15 kpps inline detection
Energy Efficiency	67% reduction compared to standard x86 IDS

Tools I Trust

Recruiter capability map.

Languages

Java 100%
Python 95%
Dart 90%
SQL 85%
JavaScript 82%
C / C++ 80%
Bash Scripting 80%

Backend & Systems

REST API Design 95%
Clean Architecture 90%
Object-Oriented Design 90%
Database Indexes 85%
Flask Core 82%

Networking & Sec

TCP/IP & Handshakes 95%
Vulnerability Audits 90%
Intrusion Detection 90%
Ethical Hacking 88%
Wireshark / iptables 85%

Linux & DevOps

Debian / Arch Admin 95%
Docker Containers 90%
Git / GitHub Actions 92%
Server Hardening 88%
CI/CD Workflows 80%

AI & Models

Deep Learning 85%
Model Quantization 82%
Edge Inference TFLite 80%
AI Agent Orchestration 90%

Mobile & Frontend

Flutter 95%
Riverpod State 90%
Material Design 85%
Typography & Grid 82%

Engineering Ethos

"Being weak is nothing to be ashamed of… Staying weak is!"

— Fuegoleon Vermillion

I believe in thinking from first principles — understanding data flow, control paths, and failure modes before writing a single line of code. I value performance, clarity, and maintainability over novelty. Given sufficient time and access, I focus on understanding the problem deeply and then building reliable, maintainable solutions.

Gallery

Through my lens.

Affinity

Creatures I adore.

Field Notes

Words I keep.

Collaboration

Let's build something reliable.

I'm always open to discussing new projects, backend systems, security research, internships, or engineering discussions.

vyasdevgna@gmail.com github.com/vyas-devgna linkedin.com/in/devgna-vyas