Computer Science Graduate · Full-Stack Engineer · AI Builder
Building intelligent
products that scale.
I design and build production-ready AI systems, full-stack applications, and scalable software — from research to real-world deployment.
[ NEURAL ARCHITECTURE ]
[ AUTONOMOUS SYSTEMS ]
UNIT
AI AGENT
STATUS
ONLINE
Systems built
to solve
real problems
SentinelAI
Hybrid Edge–Cloud Violence Detection Architecture
LendGraph
Private credit infrastructure for automated loan management and investor intelligence
Fitness Tracking API
Production-grade REST API with RBAC, HATEOAS, and ETag caching
RegulaPilot
AI-powered compliance workflow system with a real backend and structured LLM pipeline
Café Digital Growth & Operations Platform
End-to-end system for customer engagement, retention, and business operations
Explore the Architecture
Step inside the systems behind my projects — from user action to backend response.
Click any node to explore its role in the system
RESTful API with JWT authentication, RBAC middleware, and a layered backend architecture built on Koa.js and MySQL.
Vue SPA Client
Single-page application that consumes the REST API. Sends Bearer tokens with every authenticated request and handles token-refresh logic on 401 responses.
- RESTful API following Richardson Maturity Model Level 3
- JWT authentication with role and scope-based access control
- ETag support for efficient conditional GET requests
- Zod validation and strict layered backend architecture
- Comprehensive endpoint testing with Jest and Supertest
Featured Case Study
Engineering in depth
SentinelAI
Hybrid Edge–Cloud Violence Detection Architecture
The Problem
Modern fitness applications demand APIs that go beyond basic CRUD — handling complex user hierarchies, enforcing fine-grained access control, and remaining self-documenting at runtime without a separate documentation step.
Architecture Layers
Camera Feed — Continuous CCTV frame capture feeding into the edge processing pipeline at full frame rate.
Edge AI — MobileNetV2 — Lightweight CNN performing real-time violence classification on-device with minimal compute requirements.
Suspicion Threshold — Confidence scoring gate that determines whether a frame warrants escalation to cloud verification.
Vision-Language Model — OpenAI Vision API performs multimodal semantic analysis of suspicious frames for contextual verification.
Decision Engine — Final classification combining edge confidence scores and VLM verification for reliable event detection.
Logging & Analysis — Structured event logging, audit trail, and analytics dashboard for monitoring detection performance.
Key Engineering Decisions
- JWT with refresh token rotation — zero long-lived secrets in circulation
- RBAC permission matrix — O(1) authorization checks at every endpoint
- ETag caching — ~60% bandwidth reduction on read-heavy workloads
- HATEOAS responses — API is self-documenting at runtime
- Integration test suite — real database, real assertions, no mocks at data layer
Tech Stack
Read the full case study
Architecture, security, testing, and outcomes
Expertise
Skills & Technologies
Backend
Frontend
Mobile
AI / ML
Cloud & Infra
Tools
About
The engineer
behind the work
I build systems with architecture in mind first — thinking about security, scalability, and maintainability before a single line is written.
My focus spans the full stack: from designing secure REST APIs with RBAC and HATEOAS, to building cross-platform mobile apps with offline-first architecture, to training deep learning models for video classification. The common thread is engineering rigour and a preference for systems that actually hold up under production conditions.
I write code to be read by the next engineer — tested, documented at the right level, and structured so that change doesn't require archaeological excavation.
Engineering Principles
Architecture first
Design the system before writing the code.
Security by default
Baked in from day one — not retrofitted.
Test what matters
Integration over mocks; confidence over coverage theatre.
Clean over clever
The next engineer should thank you, not curse you.
Ready to build
Let's build something impactful
I'm currently available for freelance work, contract roles, and full-time positions. If you have a technical challenge worth solving, let's talk.
Contact
Start a conversation
Have a project in mind, a role to fill, or just want to connect? Reach out directly.