Skip to content

Full index

38 additional projects

  • Aflon Mark

    Aflon Mark records daily staff attendance from a mobile QR scanner and RFID card readers, with an admin dashboard for staff, schedules and reports. I built a Next.js API with 18 route files and 21 handlers, 3 MongoDB models, 7 admin screens, CSV exports, and an Expo app with an offline scan queue. All 31 commits across both repos are mine, with 144 automated tests.

    • Full Stack Software EngineerSenior
    • Expo
    • React Native
    • Next.js
    • MongoDB
    • AsyncStorage
  • Tayari Tutor Matching

    Tayari Tutor Matching recommends tutors to language learners on the Tayari marketplace. I first trained a Keras scoring model on 300 synthetic student-tutor pairs, then rebuilt matching as semantic search: tutor profiles are embedded locally with Ollama nomic-embed-text in batches of 16 and stored in LanceDB, and learner requests are searched with SQL-style filters for budget, subjects and time-slot overlap. I then ported it into the team marketplace as an Express tutor-matching service.

    • AI/ML EngineerMid Level
    • TypeScript
    • LangChain
    • Ollama
    • LanceDB
    • Express
    • Vector search
  • Cassava Disease Classifier

    I built a cassava leaf disease classifier that uses a frozen ImageNet ResNet50 as a feature extractor and a 100-tree Random Forest on the pooled features, trained on 7,508 leaf images across 5 classes. The hybrid reached 0.89 test accuracy against 0.59 for a Random Forest on raw pixels, and I serve it through a Streamlit app with upload, camera and sample-image tabs.

    • AI/ML EngineerMid Level
    • Academic Researcher
    • Python
    • ResNet50
    • Random Forest
    • Streamlit
  • BrainWave

    BrainWave is an online learning platform where students enrol in tutor-led courses, join live classes and use an AI assistant. I built the TypeScript Express backend with 88 routes across 16 route modules and 15 Mongoose models, Stripe and wallet payments, Jitsi JaaS tokens for live sessions and Socket.IO messaging. Its AI service routes chat to Gemini, OpenAI or Anthropic from a MongoDB setting and adds context from 6 skill files. I also built the student app and a separate admin console.

    • AI/ML EngineerMid Level
    • Full Stack Software EngineerMid Level
    • React
    • Node.js
    • Express
    • MongoDB
    • LangChain
  • Tayari AI Course Media

    Tayari AI Course Media generates curriculum, narration and video for tutors building language courses on the Tayari marketplace. Inside the team's Express backend I wrote provider classes for ElevenLabs and Google text-to-speech and for Veo 2 video. The Veo provider renders up to 8 clips from distinct, culturally specific scene prompts, joins them with FFmpeg and merges the narration track. Seven instructor-only routes expose outline, lesson, exercise, audio, video, metadata and lecture generation.

    • AI/ML EngineerMid Level
    • Node.js
    • ElevenLabs
    • Google Text-to-Speech
    • Veo
    • FFmpeg
  • AFLON Smart Campus Concierge

    The Smart Campus Concierge lets staff upload school handbooks and policies and ask questions by text or voice. I built it as a client-side React app in about 1,100 lines: PDF.js extracts text, 4,000-character chunks with 1,000-character overlap are stored in IndexedDB, keyword scoring picks up to 30 passages plus neighbours, and Gemini 2.5 Pro streams the answer. Voice uses the Gemini Live native-audio model.

    • AI/ML EngineerMid Level
    • React
    • PDF.js
    • IndexedDB
    • Gemini
  • Hospital AI Translation

    I built a translation prototype for doctor-patient conversations: a Flask API serving three fine-tuned MarianMT models that I published on Hugging Face (Yoruba to English, Nupe to English, Nupe to Yoruba), and a Flutter client that posts text to the API. The repo includes a 144-sentence trilingual clinical phrase dataset.

    • AI/ML EngineerMid Level
    • Translation
    • Flask
    • Flutter
  • Open TutorAI Contribution

    In a fork of Open TutorAI I made 3 commits out of 383: a configurable model allowlist for OpenAI-compatible providers, a Markdown-stripping utility so the avatar's text-to-speech reads clean prose, a FastAPI adapter that lets the app generate images through NVIDIA's API, dev service scripts, and rebranding to Aflon Tutor AI across 42 files.

    • AI/ML EngineerMid Level
    • Open Source
    • Model Configuration
    • Text to Speech
  • Sunbeth Fuel Forecourt

    Sunbeth Fuel Forecourt is a forecourt controller prototype for a Nigerian fuel station that releases fuel only after an OPay POS payment is confirmed. I wrote the ESP32 firmware, a 7-state machine that drives a solenoid relay and counts flow-meter pulses on an interrupt. I also built the TypeScript Express gateway with 9 REST endpoints, HMAC-SHA512 request signing, a WebSocket channel to dispensers and a 6-table SQLite database, plus a React SCADA dashboard and simulators for bench testing.

    • Full Stack Software EngineerMid Level
    • React
    • Express
    • ESP32
    • WebSockets
    • SQLite
    • OPay
  • Oduduwa Platform

    The Oduduwa platform runs the membership, dues, events and publications sites for two related organisations from one Express API. I built a tenant resolver that picks per-request Sequelize models from two MySQL databases with shared user identity, 80 API routes across 14 route files, 13 model definitions, Paystack dues payments, a 276-line MongoDB-to-MySQL migration script, and two React front ends with 29 pages each.

    • Full Stack Software EngineerMid Level
    • Express
    • MySQL
    • Sequelize
    • Multitenancy
  • Beyond Emotions

    Beyond Emotions is a website for a counselling and devotional ministry, with a blog, devotionals, services and a contact path. I built it on TanStack Start with an embedded Sanity Studio backed by 20 schema types, and 29 server functions that validate input with Zod and store contact forms, newsletter sign-ups and visitor chat in 4 MongoDB models. Nodemailer sends email notifications, and a protected dashboard lets the owner read submissions and reply to chats.

    • Full Stack Software EngineerMid Level
    • TanStack Start
    • Sanity
    • MongoDB
  • KBM Website and Sprint Registration

    This is the public website for Know the Blocks Maven plus the separate sign-up app for its paid Sprint. The site runs on Next.js with a Sanity page builder of 32 section types, 17 public pages and 10 API routes for contact, newsletter, donations and Paystack. The sprint app takes payment for three plans and records it in MongoDB with an atomic upsert so the callback and webhook send confirmation emails through Resend only once.

    • Full Stack Software EngineerMid Level
    • Next.js
    • Sanity
    • Paystack
    • Resend
  • Sakai Deployment and Customization

    This work packages the upstream Sakai 23 LMS for repeatable deployment and adapts it for multiple universities on aflon.com.ng. I built a multi-stage Docker image that compiles Sakai with Maven onto Tomcat 9, a manage.sh CLI with 16 commands, and Nginx configs for 7 sites. In sakai-project-dev I patched the login tool so each tenant host gets its own skin and title, and added a Faculties tool (8 Java classes, 6 Velocity views) for faculty, department, level and semester browsing.

    • Full Stack Software EngineerMid Level
    • Sakai
    • Docker
    • Java
  • Moodle Installer

    MoodleDeploy is a Bash tool that installs or upgrades Moodle 4.1 to 5.1 on Linux and Windows (Git Bash, MSYS2 or WSL). It covers four cases: a fresh server, an existing server, cPanel hosting, and generating a GitHub Actions pipeline. I wrote all 4,171 lines of shell across 14 scripts, with per-version PHP and database requirement checks and an upgrade manager. 156 Bats test cases cover it.

    • Full Stack Software EngineerMid Level
    • Bash
    • Moodle
    • Deployment
  • Portfolio CMS

    I built this portfolio as a Next.js 16 app with an embedded Sanity Studio at /cms, 7 content schemas, tag-based revalidation from a signed Sanity webhook, and a portfolio assistant that calls an NVIDIA-hosted model with 3 retrieval tools over my project records. It also syncs Kaggle notebooks into Sanity. All 74 commits are mine.

    • Full Stack Software EngineerMid Level
    • Next.js
    • Sanity
    • TypeScript
  • Tayari Marketplace

    Tayari Marketplace is a team platform where learners find tutors for African languages and tutors publish courses. Across the two team repos I made 20 commits: the tutor-matching service and onboarding recommendations in v1, and in v2 the AI course generation and media providers, OTP email verification with configurable SMTP, Vercel SPA routing and redesigned Home, About, Contact, Login and Signup pages. The v2 backend has 57 Express routes and 12 Mongoose models, built with teammates.

    • Full Stack Software EngineerMid Level
    • Marketplace
    • Onboarding
    • Course Media
  • Magazine Archive

    On a team newsroom site (4 of 45 commits mine), I integrated Sanity CMS: 9 schema types including articles, events, magazines and four page singletons, an embedded Studio, a 489-line seed script, and magazine listing and detail pages with an inline PDF viewer and download. I also added an event registration modal.

    • Full Stack Software EngineerMid Level
    • Next.js
    • Sanity
    • Publishing
  • Mandura Storefront

    I built a single-page storefront for Mandura, a herbal tonic, with a GSAP ScrollTrigger story sequence, a CSS 3D bottle viewer, For Her and For Him variants, and a cart drawer. Products, prices and stock load from the Shopify Storefront GraphQL API, and checkout redirects to Shopify's hosted checkout via cartCreate.

    • Full Stack Software EngineerMid Level
    • React
    • Express
    • Shopify
  • School Result Management

    I forked a plain-PHP student result project, moved it into legacy/, and rebuilt it in Laravel 13 across 5 commits: 26 Eloquent models, 30 migrations, 28 controllers, 7 domain services for grading, score sheets, report cards and result-checker PINs, and 33 tests. Teachers enter scores on validated sheets and admins issue printable PIN batches.

    • Full Stack Software EngineerMid Level
    • Laravel
    • Migration
    • School Records
  • Wamido Website

    The Wamido website is the public site for the Wamido mentorship organisation. I built it with Next.js 15 and React 19, with a Sanity Studio embedded at /studio so staff can edit 9 page and content schemas, and a migration script that seeded the CMS from JSON. It has 9 public pages, a Resend-backed contact API, and 8 Vitest test files covering responsive hooks and the navbar.

    • Full Stack Software EngineerMid Level
    • React
    • Headless CMS
    • SEO
  • ACAIR Website

    The ACAIR website presents the institute's programmes, publications, team and book catalogue, and is the entry point to the student portal. I built it in React and TypeScript with Vite, 17 pages including 6 programme pages, and compressed media to 27 WebP assets. The code started as a Lovable scaffold; I made 42 of the 46 commits on visionary-web-solutions.

    • Full Stack Software EngineerMid Level
    • Next.js
    • Node.js
    • MongoDB
  • Fake Naira Detector

    I built a small Keras CNN that classifies Naira notes as genuine or counterfeit from 64x64 RGB images. I generated the counterfeit class by applying random colormaps, contrast, brightness, noise and blur to genuine notes, giving 576 images across four denominations. A Streamlit app shows training curves and the classification report and runs predictions on uploads.

    • AI/ML EngineerJunior
    • Keras
    • Image Classification
    • Synthetic Data
  • Human Activity Classification

    I built a pipeline that flattens per-segment wearable sensor files, standardises them, reduces them with PCA to 50 components, and compares Decision Tree, Random Forest and SVC classifiers with timing and confusion matrices across 19 activities. The saved Random Forest, scaler and PCA are served in a Streamlit app that classifies an uploaded sensor .txt file.

    • AI/ML EngineerJunior
    • Academic Researcher
    • Sensor Data
    • PCA
    • Scikit-learn
  • Tesla Random Forest Predictor

    I trained a 100-tree Random Forest regressor on daily TSLA OHLCV data from 2015 to 2024 pulled with yfinance to predict the next day's close, and built a Streamlit app where users enter today's open, high, low, close and volume. The app recomputes MSE, RMSE, MAE and R² live and falls back to stored metrics when Yahoo Finance rate-limits.

    • AI/ML EngineerJunior
    • Quant TraderJunior
    • Random Forest
    • Time Series
    • Streamlit
  • COCO Object Detection Demo

    I wrote OpenCV scripts that load a pretrained SSD MobileNet v2 COCO model through cv2.dnn, run it on webcam frames or still images at 300x300, and draw labelled boxes above a 0.25 confidence threshold. A second script runs OpenCV's ResNet-10 SSD face detector on a video stream. The pretrained weights come from TensorFlow and OpenCV, not my own training.

    • AI/ML EngineerJunior
    • OpenCV
    • COCO
    • Object Detection
  • Learning Assistant AI

    I built an Expo React Native study assistant with three tabs: a Gemini 2.5 Flash chat, a visual tab that sends camera or gallery photos of notes to Gemini vision, and spoken answers generated by Gemini's TTS model with the Aoede voice. Responses are cleaned of Markdown before speech, and a custom math helper renders LaTeX as Unicode and 2D fractions.

    • AI/ML EngineerJunior
    • Expo
    • Gemini
    • Speech
  • Smart Buddy AI

    I built a browser voice tutor for junior secondary school learners that streams microphone audio at 16 kHz to the Gemini Live native-audio model and plays back 24 kHz spoken replies with live transcripts. Learners pick one of 6 JSS subjects with suggested topics, and a curriculum-grounded system prompt shapes the answers.

    • AI/ML EngineerJunior
    • React
    • Vite
    • Gemini
  • University Student Success Engine

    The Student Success Engine is a prototype that classifies students as high risk, moderate risk or on track from attendance, grades, assignments and LMS logins. I built it in React with about 1,450 lines of TypeScript: Gemini 2.5 Flash returns a schema-enforced JSON array with risk score, root cause and intervention, and a voice advisor uses the Gemini Live audio model. Results export to CSV.

    • AI/ML EngineerJunior
    • React
    • Gemini
    • Structured Output
  • Attendance System

    I built a lecture attendance system: an Express 5 and MongoDB API with JWT authentication, 4 Mongoose models and 15 endpoints for lecturers, students, lectures, attendance marking and per-student reports, and a React 19 and TypeScript client with 11 pages behind protected routes. A compound unique index prevents duplicate attendance records per lecture.

    • Full Stack Software EngineerJunior
    • React
    • Express
    • MongoDB
  • Shopcart Ecommerce Recreation

    I recreated the Shopcart storefront as a React client and an Express 5 API on Node's built-in SQLite, with 18 API routes and 8 tables. It has scrypt password hashing, SQLite-backed sessions with CSRF tokens, guest carts that merge on sign-in, and Paystack hosted checkout confirmed by both return-URL verification and signed webhooks. Orders get timelines and printable receipts.

    • Full Stack Software EngineerJunior
    • SQLite
    • Sessions
    • Checkout
  • iDeliverySell

    I built an early (2023) Express app with server-rendered EJS views where delivery applicants submit their details and upload documents, and admins sign up, sign in with sessions, and review applicants on a dashboard. Data is stored in MongoDB and uploads go through Multer disk storage. It has 11 routes and 14 EJS views.

    • Full Stack Software EngineerJunior
    • Express
    • EJS
    • Sessions
  • Voting DApp

    I wrote a Solidity voting contract where the owner registers candidates and voters, only registered voters aged 18 or over can vote, and each can vote once. I deployed it with Truffle to a local chain and built a React front end that reads and writes the contract through Thirdweb hooks.

    • Full Stack Software EngineerJunior
    • Solidity
    • Thirdweb
    • Web3
  • Weather Forecast App

    I built a React dashboard and Flask API that take location, crop and current weather readings, predict the next readings with a Random Forest regressor, and classify whether conditions suit planting, fertilizing or harvesting with a Random Forest classifier. It covers 5 Nigerian cities and 5 crops, and earlier per-city models are kept in the repo.

    • Full Stack Software EngineerJunior
    • React
    • API Integration
  • Food Delivery Mobile App

    I built an Expo React Native food ordering prototype with Firebase email and password authentication persisted in AsyncStorage, a menu, food detail and cart flow managed in React context, and an order tracking screen that animates a courier marker along a polyline on Google Maps between a restaurant and the customer near LAUTECH.

    • Full Stack Software EngineerJunior
    • Expo
    • Firebase
    • Location
  • ACAIR Facial Recognition Attendance

    Facial recognition attendance application using TensorFlow and OpenCV for student verification.

    • AI/ML EngineerJunior
    • TensorFlow
    • OpenCV
    • Computer Vision
  • ACAIR School Management Mobile App

    School management app built for mobile with React Native and Expo.

    • Full Stack Software EngineerJunior
    • React Native
    • Expo
    • Mobile
  • Cashew Disease Prediction System

    A Kaggle research notebook exploring machine learning for cashew crop disease prediction.

    • AI/ML Engineer
    • Academic Researcher
    • Machine learning
    • Plant disease prediction
    • Kaggle
  • Real Time Odds Adjustment for Football

    A Kaggle research notebook exploring how new match information affects in-play probabilities and estimated odds.

    • AI/ML Engineer
    • Kaggle
    • Probability
    • Sports Analytics

Portfolio assistant

Gabriel’s projects, experience, and research