John Jomi
Selected Work

Projects

A focused set of portfolio projects spanning enterprise platforms, AI applications, scientific computing, and cybersecurity research tooling.

Featured Project/01

IEEE Student Branch ERP

A full-scale enterprise resource planning platform built for IEEE Student Branch operations. The system streamlines society management, finance, event coordination, announcements, authentication, cloud storage, and AI-assisted workflows into a single platform.

Highlight Features

•Society Management•Event Management•Financial Tracking•Role-Based Authentication•AI Assistant•Azure Cloud Storage•REST APIs•Responsive Dashboard

Technology Stack

Next.js·React·TypeScript·Node.js·Express·Prisma·PostgreSQL·Azure

Project Details

Status: In Development

Project Links:

GitHubView Case Study
In Development

ZeroProof — Zero-Knowledge Authentication

Web-based zero-knowledge proof system that lets a user prove knowledge of a secret to a server without the secret, or anything derived from it, ever crossing the network.

TypeScript · Next.js · React · Cryptography

GitHubCase Study
Completed

HealthAssist — Health Information Navigator

Conversational assistant that helps users prepare for a doctor's appointment by organizing symptoms, flagging high-risk situations for escalation, and formulating questions to ask.

Next.js · TypeScript · Google Gemini · LLM

GitHubCase Study
In Development

LearnSmart — AI Learning Platform

Production-oriented AI learning platform built as a NestJS modular monolith, spanning auth, documents, courses, lessons, quizzes, AI generation and progress tracking.

NestJS · TypeScript · PostgreSQL · Prisma

GitHubCase Study
In Development

AI Chatbot

Conversational AI assistant built using modern language models and prompt engineering techniques to provide intelligent responses and workflow assistance.

OpenAI API · Next.js · Node.js · LLM

GitHubDemo
Featured Project/02

Eventra — Intelligent Event Management Platform

Enterprise-grade platform that automates the full lifecycle of complex events, from ticketing and payments to AI-assisted matchmaking and attendee communication.

Highlight Features

•Event Lifecycle Automation•AI Matchmaking•Vector Recommendations•Ticketing & Payments•Communication Hub•Clerk Authentication•Server Actions•pgvector Search

Technology Stack

Next.js·TypeScript·PostgreSQL·pgvector·Drizzle ORM·Clerk·Google Gemini·Vector Search

Project Details

Status: In Development

Project Links:

GitHubView Case Study
Completed

FraudGraph — Credit Card Fraud Detection with GNNs

Full-stack fraud detection system that models card transactions as a graph and classifies them with a Graph Convolutional Network, served through FastAPI and explored in a React dashboard.

Python · PyTorch · Graph Neural Networks · FastAPI

GitHubCase Study
Completed

SupportDesk — Ticket Management System

Full-stack MERN support ticket platform with assignment workflows, customizable statuses, priority levels, department routing, internal notes, ticket merging and audit history.

React · Node.js · Express · MongoDB

GitHubCase Study
Completed

NASA Seismic Noise Filtering System

Scientific computing application that retrieves, processes and filters seismic recordings from NASA's IRIS dataset, with specialized support for Martian data from the InSight mission.

Python · ObsPy · Streamlit · Scientific Computing

GitHubCase Study
Featured Project/03

Tycoon 2FA — AiTM Phishing Detection

Research prototype that detects Tycoon 2FA and adversary-in-the-middle phishing email by combining four independent detection layers into a single explainable risk verdict.

Highlight Features

•Header & Domain Intelligence•Redirect Chain Analysis•QR & Base64 URL Extraction•NLP Body Classification•Threat Intel Correlation•Composite Risk Scoring•Explainable Evidence•Read-Only Gmail Ingest

Technology Stack

Python·Security Research·Threat Detection·NLP·Machine Learning·Email Security

Project Details

Status: In Development

Project Links:

GitHubView Case Study
Completed

Tomato Freshness Classification with Explainable AI

Full-stack ML system that classifies tomato freshness from gas-sensor and environmental readings and explains every prediction with feature importance, LIME and SHAP.

Python · Scikit-learn · XGBoost · SHAP

GitHubCase Study
Completed

Email AI Classifier

Gmail classification system that uses Google Apps Script and Gemini batch inference to apply hierarchical labels automatically, surfaced through a companion Safari extension.

Google Apps Script · Gemini API · JavaScript · Safari Extension

GitHubCase Study
Completed

HRM Research Engineering

Independent research and engineering work on the Hierarchical Reasoning Model, porting the reference implementation to Apple Silicon and hardening its evaluation pipeline.

Python · PyTorch · Research · Apple Silicon

GitHubRepository
Featured Project/04

SkinAIBot — Medical AI Diagnosis Platform

Medical AI platform for skin disease diagnosis and analysis, built on a separated two-team architecture that splits the full-stack application from the AI/ML service.

Highlight Features

•Skin Lesion Classification•Two-Team Architecture•Separated AI Service•JWT Authentication•Alembic Migrations•Dockerized Stack•FastAPI Backend•PostgreSQL Storage

Technology Stack

Python·FastAPI·Next.js·PostgreSQL·Docker·Alembic·Medical AI·Deep Learning

Project Details

Status: In Development

Project Links:

GitHubView Case Study
Completed

Music Recommender — AI-Powered Spotify Recommendations

Full-stack application that authenticates users through Spotify OAuth, surfaces their top tracks and artists, and generates personalized song recommendations with Google Gemini.

Next.js · TypeScript · FastAPI · Python

GitHubCase Study
Completed

Email Form Collector

Form builder that sends surveys through a connected Gmail account via OAuth and collects responses on an authenticated dashboard, with encrypted credentials and rate limiting.

Python · FastAPI · SQLAlchemy · Alembic

GitHubCase Study
Completed

Stock Price Prediction with Random Forest

Regression model that predicts stock closing prices from historical Yahoo Finance data using Random Forest, evaluated with MAE, MSE and R² against actual prices.

Python · Scikit-learn · Random Forest · Machine Learning

GitHubCase Study