Building intelligent
systems that scale

The Challenge

Enterprise organizations struggle with ISO compliance documentation—a process that's traditionally time-consuming, error-prone, and resource-intensive. Manual documentation workflows lead to inconsistencies, delays in certification, and significant operational overhead.

The Solution

As Project Lead and Backend Developer, I architected and developed ISO SPACE—an enterprise-grade automation platform combining intelligent backend logic with Firebase API integration to deliver structured, maintainable compliance documentation.

• Scalable Backend Architecture: Modular, maintainable code structure supporting multiple ISO standards
• Firebase API Integration: Real-time data sync and cloud-based document management
• Automated Workflow Engine: Intelligent automation generating compliance reports and managing audit trails
• Cross-Platform Deployment: Flutter + Google Cloud for web and mobile access

Technical Stack

Impact & Results

• ~40% Time Reduction: Automated workflows dramatically cut documentation preparation time
• 100+ Enterprise Clients: Deployed across numerous organizations requiring ISO compliance
• Dual ISO Certification: Platform achieved ISO/IEC 27001 and ISO/IEC 20000-1 certifications
• Audit-Ready Documentation: 100% compliance accuracy through automated validation

The Challenge

APPEKNAS needed a reliable, nationwide system to manage and verify professional certifications across Indonesia's 38 provinces with distributed data, verification integrity, and instant validation capabilities.

The Solution

• QR Verification System: Secure QR code generation for instant certificate authentication from any device
• Multi-Tier Admin System: Hierarchical access controls for national, provincial, and organizational roles
• Database Architecture: Scalable MySQL schema handling millions of records with optimized performance
• API Integration: RESTful APIs for third-party verification and mobile app integration

Technical Stack

Impact & Scale

• National Coverage: Deployed across all 38 Indonesian provinces as the official APPEKNAS verification platform
• Zero Fraud: No certificate fraud incidents since deployment via secure validation mechanisms
• Partnership Expansion: Facilitated international partnerships including Vietnam Construction Company collaboration

Research Problem

Detecting DoS and DDoS attacks remains critical as evolving attack patterns defeat traditional signature-based systems and conventional ML models that struggle with temporal dependencies in network traffic.

My Role & Contribution

• Designed and implemented a hybrid GRU–NTM deep learning architecture for intrusion detection
• Engineered the full data preprocessing and sequence modeling pipeline
• Developed sliding-window temporal representations for network traffic
• Led model training, evaluation, and performance analysis as first author

Architecture Overview

The model combines GRU's sequential learning with Neural Turing Machine's external memory, enabling both short-term pattern recognition and long-term contextual awareness.

Key Results

• Achieved approximately 99% classification accuracy across benchmark datasets
• Strong generalization on UNSW-NB15 and BoT-IoT datasets
• Consistent ROC-AUC performance across all attack classes

Publication Information

• Conference: IEEE ICOSEC 2024
• DOI: 10.1109/ICOSEC61587.2024.10722438
• Publisher: IEEE Xplore Digital Library

Research Motivation

Urban parking challenges cost drivers time and contribute to congestion and emissions. Traditional systems lack intelligent automation and real-time availability information critical for modern urban infrastructure.

System Architecture

• License Plate Recognition: Deep learning OCR for accurate vehicle identification across varying conditions
• IoT Sensor Integration: Raspberry Pi edge computing nodes for distributed parking space monitoring
• Real-time Processing: Optimized inference pipeline for low-latency detection
• Cloud Integration: Centralized management for multi-location monitoring and analytics

Technology Stack

Results & Outcomes

• Robust license plate detection across diverse vehicle types and conditions
• Low-latency processing suitable for immediate parking status updates
• Scalable architecture supporting multi-facility expansion
• IEEE publication at ICPCT 2025 contributing to smart city research

Innovation Overview

An AI-enabled beverage preparation and dispensing system that customizes beverages based on user health data and real-time physiological state detection, integrating artificial intelligence, embedded systems, robotic processing mechanisms, and automated ingredient management.

Core Technical Innovation

• AI-based imaging unit for facial fatigue detection and physiological assessment
• Health-driven adaptive recipe selection via integrated microcontroller and database
• Real-time ingredient substitution based on user dietary restrictions
• End-to-end automation from raw ingredient processing to final beverage dispensing

Patent Details

• Inventors: Sayani Roy, Dr. Janaki Kandasamy, Caroline Panggabean, Yelicherla Tejashree
• Application Number: 202541042823 A
• Jurisdiction: India
• Filing Date: May 2, 2025
• Publication Date: May 30, 2025
• Status: Published Application

Official Documents

• View Published Patent Specification
• View Technical Drawings