Kiumbura N.
Githinji

Overview

KNGCuts is a full-stack barbershop booking platform built for a real, active business operating at Monmouth University. The app lets clients browse services, book appointments online, and pay seamlessly — while giving the owner a complete admin dashboard to manage the business.

Booking

Admin Dashboard

Gallery

Key Features

Services & Pricing

Tech Stack

Overview

As the founding president of Monmouth University's IEEE/ACM chapter, I designed and built the official website from scratch. It serves as the central hub for the organization — handling event promotion, committee info, membership recruitment, and fundraising for official IEEE international recognition.

Homepage

Events Calendar

Activities & Programs

Organization Structure

Designed a 5-committee structure to scale the chapter's operations:

Events Calendar

Impact & Goals

Overview

ATLAS is a production-grade educational platform purpose-built for teaching automata theory at the university level. It combines structured course content spanning 9 modules (from DFA through P vs NP), interactive automaton building and simulation tools, study aids, an AI tutor chatbot, and a full instructor management suite with gradebook and roster tracking.

Course Center

AI Tutor

Quiz Mode

Platformer Game

Game Menu

Course Curriculum

9 comprehensive modules covering all of theoretical computer science:

Interactive Tools

Instructor Suite

Student Experience

Demo Videos

The Challenge

The U.S. has over 12,000 miles of coastline to monitor, and roughly 95% of global trade moves by sea. The Strait of Hormuz alone sees ~21 million barrels of oil transit every day — a chokepoint adversaries probe through AIS spoofing, dark transits, fast-boat swarms, and mine-laying. Cloud-dependent surveillance is too slow, too expensive, and useless in denied or degraded environments. There was no edge-first alternative — so we built one in 48 hours.

Overview

Hormuz Watch is a tactical maritime domain awareness system designed to detect anomalies and threats in the Strait of Hormuz through edge-based machine learning, with zero reliance on cloud infrastructure. Built in 48 hours at the Philly 250 Maritime Defense Hackathon (March 27–29, 2026) at the 23rd Street Armory in Philadelphia, the system detects suspicious vessel behavior via an ML anomaly-scoring ensemble running on a Raspberry Pi 5 with sub-15ms latency, relays encrypted alerts through an ESP32 mesh network, and performs local LLM-powered threat analysis on a Jetson Orin Nano — all for under $450 in hardware.

Four Pillars — Detect · Relay · Analyze · Act

Every layer runs at the edge with graceful degradation — no single point of failure, no cloud.

Hardware in Action

The full sensor-to-commander stack — built and demoed live over 48 hours at the 23rd Street Armory.

Full deployment — Jetson Orin Nano, Raspberry Pi 5 edge nodes, and ESP32 mesh radios.

Edge mesh node — a Raspberry Pi compute unit wired to an ESP32 relay radio.

Live system — real-time anomaly detection logs beside the operator dashboard.

The build bench — edge nodes, mesh radios, and dashboards mid-hackathon.

System Architecture

The ML Engine

A two-model ensemble flags anomalous vessels across an 18-feature space, trained on real USCG AIS data and benchmarked against 82 vessel-type baselines.

Validated Incident Replays

Tested against 5 real-world maritime incidents with 100% recall:

Performance & Validation

Deployment Economics

Built to scale across chokepoints — every node is cheap, disposable, and independently replaceable.

Tech Stack

Team