Open to opportunities

Kijumbe James

@kijumbejames

Message

I build production web apps and on-device AI systems for real-world problems.

Tanzania

Message

What I'm looking for

I'm looking to build and ship practical full-stack, data, and AI products, especially systems that solve real-world problems. I value independent work, clear communication across time zones, and opportunities to grow through hands-on product development.

I built LeafScan TZ, an offline Android app for cassava and maize disease detection, achieving 95.6% accuracy across 20 crop-disease classes. I trained EfficientNetLite0 and MobileNetV2 models, resolved out-of-distribution misclassification through Outlier Exposure retraining, and optimized models to TensorFlow Lite as small as 2.75MB.

Across independent projects, I've designed, built, and deployed 10+ full-stack and API products for fintech, healthcare, inventory, and note-taking use cases using Flask, Node.js/Express, React, Next.js, MongoDB, and MySQL. I've also built AI chatbot interfaces and deployed cloud-optimized applications on Vercel and Render.

At ChinaBoda Co. Ltd, I used AI tools and prompt engineering to improve support content and automate routine customer inquiries. My data operations experience at Muhimbili National Hospital strengthened my attention to accurate, reliable record-keeping in a high-volume environment.

Experience

Work history, roles, and key accomplishments

Current

Full-Stack & AI Developer

Current

Independent Projects

Jan 2024 - Present (2 years 7 months)

Designed and trained EfficientNetLite0 and MobileNetV2 models for LeafScan TZ, an offline Android app for crop disease detection, achieving 95.6% accuracy and 0.93 Macro F1 across 20 classes. Built and deployed 10+ full-stack and API projects using Flask, Node.js, and React.