Now building Targuma, an AI-native publisher

I build AI products and ship them all the way to your pocket.

AI lets one person build what used to take a whole team. I spent seven years shipping apps used by millions at The New York Times, Intuit, Kiva and Kaiser Permanente. Now I build AI-native products end to end, from the model pipeline to the App Store.

Targuma app showing a book's English edition in the reader
Bete Tselot500K+ downloads · 4.9★
500K+downloads of Bete Tselot, an app I built solo
4.9★average rating on Google Play
+35%daily active users from the NYT feature I led
2courses authored and taught for Udacity

Shipped at

The New York TimesIntuitKivaKaiser PermanenteUdacity

Selected work

Products I've built and shipped.

Real products with real users, not demos. I design, build and ship each one myself, with AI agents doing what a team used to.

Targuma app icon
TargumaFounder · AI-native publishing · 2026

Books that cross languages.

The Kindle store doesn't sell books in Amharic, a language read by tens of millions of people. So I built a publisher for the languages big stores leave out. AI drafts the translation, the author reviews it line by line beside the original, and nothing ships without their consent. Readers get both editions side by side, in days instead of years.

  • AI translation pipeline that drafts a whole book and aligns it paragraph by paragraph for review
  • Publishing rail that turns one manuscript into an ebook and a print-on-demand paperback
  • Web store and Flutter apps for iOS and Android, with Stripe and in-app purchases
  • Agent Abay, an AI teammate that writes code and reviews every pull request
  • LLM pipelines
  • Python
  • Flutter
  • Stripe
  • StoreKit
  • Play Billing
A book page in the original Amharic in the Targuma reader
original
Targuma home screen featuring the book Atsedu
store
The same book page as an English Targuma edition
targuma edition
Bete Tselot app icon
Bete TselotSolo build · iOS & Android

Half a million downloads. One developer.

A daily prayer companion for the Ethiopian Orthodox community: prayers in Ge'ez, Amharic, Tigrinya, English and Afaan Oromo, streaming audio, the Bible, the church calendar and community prayer requests. I design, build and run all of it, on phones, tablets and foldables.

500K+downloads
4.9★Google Play
4.8★App Store
Bete Tselot audio prayers screen
Bete Tselot Today screen with language choices
Bete Tselot Today screen in dark mode
HalehameiOS keyboard · 2026

Type Amharic the way you say it.

A keyboard for iPhone that turns phonetic Latin typing into Ge'ez script as you go, with live suggestions, the full fidel chart and English-to-Amharic translation built right into the keyboard.

  • Swift
  • Keyboard extension
  • Transliteration
  • Translation
Halehame keyboard converting phonetic typing into Amharic
Halehame fidel chart
Halehame translating English to Amharic inside the keyboard

How I build

One builder. A team's output.

AI lets one person build what used to take a whole team. This is the workflow I use every day and teach in Udacity's Introduction to AI-SDLC. Agents do the typing. I do the judgment.

01

Frame the problem

Start from a real person with a real problem. Write the spec, the success metric and the checks that tell me when the AI is good enough.

spec.md
02

Build with agents

Claude, Cursor and Copilot write the first drafts. I own the architecture, read every diff and decide what ships.

claude · cursor · copilot
03

Review like a team

An AI reviewer checks every pull request and builds test binaries before a human merges. Quality doesn't depend on headcount.

agent-abay → PR review
04

Ship to every screen

Web, iOS and Android, with payments, analytics and store releases done properly. A product, not a demo.

web · iOS · Android

Meet my teammate

Agent Abay, the AI engineer on my team.

Abay lives in Slack. It picks up tasks, writes code, builds test binaries for iOS and Android, and reviews every pull request before I merge. Every Targuma production deploy goes through it.

Teaching & AI training

I teach engineers to build this way.

Shipping with AI is a skill, and I like passing it on. I write courses for Udacity and help make AI models better at writing code.

Udacity · Course author · 2026

Introduction to AI-SDLC

A multi-month course on using AI coding agents across the whole software lifecycle: planning, implementation, testing and review. It's the workflow on this page, written down for other engineers.

Now writing
Udacity · Instructor

Android Kotlin Developer Nanodegree

Taught modern Android to engineers around the world: Jetpack Compose, Hilt, WorkManager and testing.

4.6/5 learner rating
AI training

Making models better at code

Reviewing, grading and correcting AI-generated code. It's the feedback loop that trains better coding models, and it sharpens my own eye for what agents get wrong.

Code review · evaluation

Experience

Seven years shipping to millions.

Before AI made one person a team, I learned what production means: millions of users, real money, health data and release trains that don't wait.

  1. 2026 — nowTargumaFounderAI-native publishing for languages the big stores leave out. Translation pipeline, storefront, iOS and Android apps.
  2. 2026 — nowStealth AI startupCo-founderBuilding from zero with an AI-first workflow. I own the Python backend and data pipelines end to end.
  3. 2025 — nowUdacityCourse author & instructorAuthor of Introduction to AI-SDLC; instructor for the Android Kotlin Developer Nanodegree.
  4. 2022 — nowThe New York TimesSenior Android EngineerLed the personalized “You” tab in Jetpack Compose, driving a 35% increase in daily active users. GraphQL pagination, analytics, mentoring.
  5. 2021 — 2022KivaLead iOS & Android DeveloperLed both platforms and the full Android migration to Jetpack Compose. Shipped real-money lending and donation flows.
  6. 2020 — 2021IntuitAndroid EngineerShared libraries powering Mint and TurboTax, focused on state management, security and performance.
  7. 2019 — 2020Kaiser PermanenteAndroid EngineerPatient-facing features for KP Health Ally under strict accessibility and data-protection standards. Helped set up CI/CD.
  8. OngoingBete TselotSolo developerDesigned, built and launched on iOS and Android. 500K+ downloads, 4.9★.

About

A builder at heart. AI is the superpower.

I'm Kidus, a software engineer in Oakland. I studied computer science at UC Santa Cruz, then spent seven years shipping mobile apps at scale: news, finance, lending and health.

Then AI changed the math. With good taste and good agents, one person can now design, build and ship what used to need a team. I use that leverage to build for people the big platforms overlook, starting with my own community: Amharic readers, Ethiopian Orthodox families and the diaspora.

Away from the keyboard I read a lot of fiction and stay curious about almost everything, which is usually how the next product starts.

Kidus Elias · Oakland, CA

AI

  • Claude
  • Cursor
  • GitHub Copilot
  • Gemini API
  • Agentic workflows
  • Prompt design
  • AI code review

Mobile

  • Kotlin
  • Jetpack Compose
  • Swift
  • SwiftUI
  • Flutter
  • Dart

Backend & data

  • Python
  • FastAPI
  • GraphQL
  • REST
  • Firebase
  • Data pipelines

Shipping

  • Stripe
  • RevenueCat
  • StoreKit
  • Play Billing
  • CI/CD
  • App Store & Play releases

Contact

Building something with AI? Let's ship it.

Founders, teams and fellow builders: I'd love to hear what you're making. Email is the fastest way to reach me.

info@kiduselias.com
esc
↑↓ move↵ open