Hi, I'm Elliezer!

Role: Pipeline & on-device ML.

An iPadOS app that is built from Challenge 5 in Apple Developer Academy collab with Black Hand Gang, with a goals to help making an access for visually impaired audiences to FEEL visual artwork. In a simple way, Tactura converts 2D visual artwork into 2.5D tactile reliefs that can be 3D-printed directly and felt.

My part was the engine underneath:

I wrote the Python reference implementation of the nine-stage conversion, ported the numeric core into a Swift package, and proved the port with a differential harness against golden fixtures , zero absolute error on every deterministic stage. I also ran Depth Anything V2 Large (305M) fully on-device through Core ML, built the camera capture and the interactive RealityKit relief preview, and added the checkpointed save/resume so tuning a slider never re-runs neural inference. Exports a binary STL straight to a slicer.

Role: Developer (Presentation Layer).

An augmented-reality iOS app that teaches latte art. Pourfect uses AprilTag fiducial tracking to lock onto the cup and pitcher, then overlays a real-time, GPU-accelerated fluid simulation so you can practice pours — Heart, Tulip, Rosetta — with any liquid and get instant visual feedback. Built after pivoting from ML-based object tracking to robust, zero-training 6DoF geometry.

Role: Android Developer and Project Manager

A native Android marketplace app built entirely in Kotlin for a Indonesian perfume brand called ODD. Built with Jetpack Compose, MVVM architecture, and a clean code approach. The app features a product catalog, shopping cart, and checkout process, providing users with a seamless shopping experience. Implemented payment gateway using Xendit, and integrated email notifications for order updates. The app is designed to be responsive and user-friendly, with a focus on performance and scalability.