AI and machine learning
I teach computers to spot patterns, read documents and answer questions, so businesses can save time and catch problems early.
Open to new projects and roles
I build smart software that people actually use.
👋 Hi, I'm Joshua Chukwurah, a software engineer in Lagos. I care about the screen you tap and the systems quietly working behind it.
Scroll, swipe or use the arrow keys to move between slides.
01
Who I am
I like building things that solve real problems, whether that's an app people use every day or an AI tool doing its job behind the scenes.
I care about software that is easy to use, quick to load, careful with people's data and simple for the next developer to pick up.
What I do best
Websites and mobile apps
I take an idea from first sketch to launch, so the part you see and the part you don't work together smoothly.
- React
- TypeScript
- Flutter
Backend and data
The engine room of an app: databases, connections between services and cloud hosting that keep everything fast, reliable and secure.
- Python
- FastAPI
- PostgreSQL
- Docker
02
See it in action
Six mini demos of the problems I solve. Pick one, or let them play.
03
Things I've built
Each project shows the technical details for developers, plus a plain-English note for everyone else. Some are open source, others were built for organisations and stay private.
EHR Anomaly Detector
Healthcare AI / Private
Four PyTorch autoencoders monitor login, network, admin and patient-data-access events at 90 to 99% accuracy through a real-time FastAPI service. Deployed with Docker on AWS EC2 behind Nginx, with a self-hosted PostgreSQL database.
In plain EnglishA smoke alarm for patient records. It learns what normal looks like and raises a flag when something is off.
- PyTorch
- FastAPI
- Docker
- AWS
- Nginx
- PostgreSQL
Private project
01Multi-Provider LLM Service
Safe AI / Private
An AI microservice that integrates Gemini, Mistral and DeepSeek with automatic failover, token budgeting and compliance filtering for HIPAA, GDPR and NDPA.
In plain EnglishAn assistant that never goes offline. If one AI provider fails, another takes over, and private details are filtered out first.
- FastAPI
- Gemini
- Mistral
- DeepSeek
- Docker
Private project
02VisionETL
Document AI / Private
A six-stage clinical document pipeline: OCR, LLM extraction, validation and formatting, producing EHR-ready FHIR, HL7 and JSON from scanned records.
In plain EnglishPaper forms go in, organised digital records come out, with no retyping.
- Python
- OCR
- LLMs
- FHIR
- HL7
Private project
03NewPipeWeb
Open source
A full-featured, self-hosted YouTube frontend built on NewPipeExtractor and shipped with Docker Compose, released under GPL-3.0.
In plain EnglishYouTube without the ads, tracking or Google account, running on a server you control.
- TypeScript
- Node.js
- Docker
- GPL-3.0
BibleLLM
Language model
A lightweight character-level Transformer language model written from scratch in PyTorch, with hybrid search for exact verse lookup and training tuned for CPUs.
In plain EnglishA small AI that writes in a Biblical style and can find an exact verse. It learns on an ordinary computer, no special hardware.
- Python
- PyTorch
- FastAPI
- Apache-2.0
iOS and Android on the Web
Web simulation
A web-based simulation of both iOS and Android interfaces, written in TypeScript and released under the MIT licence.
In plain EnglishTry iPhone and Android screens in your browser without owning either phone.
- TypeScript
- MIT
04
Tools I use
You don't need to know these to work with me. They're here for the developers and hiring teams who want to know.
- 01PythonAI and backend
- 02PyTorchTrains AI models
- 03FastAPIFast web services
- 04TypeScriptSafer web code
- 05ReactInteractive sites
- 06Node.jsServer side apps
- 07TensorFlowMachine learning
- 08PostgreSQLStores data
- 09DockerRuns anywhere
- 10AWSCloud hosting
- 11FlutterMobile apps
- 12GitTracks changes