Samuel RombsSoftware EngineerContact ↗

Software Engineer

Selected work.

Android apps, connected devices, and tools for code and document retrieval.

↓

Project index 01—06

01

Document retrieval

NASA RAG

Temporary concept illustration of research documents and a planet in space.
Temporary preview

A retrieval and question-answering pipeline for NASA documents. It combines semantic and keyword search, reranks relevant passages, and returns answers with source citations.

  • Python
  • FAISS
  • BM25
  • OpenAI
Workflow overviewFrom a question to a cited answer.
  1. Ask a questionQuery rewriting
  2. Find relevant passagesSemantic + keyword search
  3. Rank the evidenceFusion + reranking
  4. Answer with sourcesCitations to NASA documents
Try NASA RAG

Start with an example

0 / 1,000

10 submissions per IP address every 10 minutes. People on the same network may share this allowance.

02

Developer tooling

Code Analysis

Temporary concept illustration of a code editor and a code review panel.
Temporary preview

A command-line agent harness for exploring codebases and making targeted changes. Structured findings, scoped file access, and verification checks keep each run focused on the task.

  • Python
  • OpenAI
  • Pydantic
Workflow overviewFrom a codebase to a checked change.
  1. Explore the repositoryScoped file access
  2. Structure the findingsCodebase analysis
  3. Make targeted changesPatch-based edits
  4. Check the workDart analysis + completion checks
Connected projects2 repositories

New Yorker Audio

A native Android player and the Python service behind it.

Android frontendPython backend

03

Android frontend

New Yorker Audio — Android App

Temporary concept illustration of a phone with an audio player and magazine artwork.
Temporary preview

A native Android app for browsing New Yorker issues and listening to narrated articles. Built with Kotlin and Jetpack Compose, it connects to the audio backend and supports background playback, saved progress, and listening history.

  • Kotlin
  • Jetpack Compose
  • Media3
  • Retrofit
Workflow overviewFrom browsing an issue to listening on the go.
  1. Browse magazine issuesIssue covers + article lists
  2. Choose an articleMetadata from the backend
  3. Play the narrationMedia3 background playback
  4. Pick up where you left offSaved progress + listening history
04

Python backend

New Yorker Audio — Backend API

Temporary concept illustration of an audio waveform connected to a backend service.
Temporary preview

A backend for an Android app that browses New Yorker issues and narrated articles. It verifies subscriber access, serves issue and article lists, and resolves narration URLs for direct playback.

  • Python
  • FastAPI
  • httpx
  • Playwright
Workflow overviewFrom a magazine issue to narrated audio.
  1. Verify subscriber accessAuthenticated HTTP session
  2. Browse magazine issuesIssue and article lists
  3. Select an articleResolve available narration
  4. Return a playback URLCached audio metadata
05

Mobile · Connected devices

WAYZN Door Opener

Temporary concept illustration of a sliding patio door and a phone controller.
Temporary preview

A Flutter app for setting up and controlling WAYZN smart sliding door openers on iOS and Android. It brings together device onboarding, remote door controls, live device status, and shared access.

  • Flutter
  • Dart
  • Firebase
  • Bluetooth LE
Workflow overviewFrom device setup to everyday door control.
  1. Discover the openerBluetooth device scanning
  2. Connect and calibrateWi-Fi setup + guided installation
  3. Control the doorDoor commands + live status
  4. Manage access and schedulesShared keys + automation
06

Mobile · Discovery

Weedmaps Android App

Temporary concept illustration of a map with location markers and a search panel.
Temporary preview

I worked across the full Weedmaps Android app for nine years. The app brings together location-based discovery, product and retailer information, community reviews, search, and filters.

  • Android
  • Kotlin
  • Jetpack Compose
  • React Native
  • Coroutines
Workflow overviewFrom local discovery to saved favorites.
  1. Choose an areaLocation-based discovery
  2. Explore nearby retailersMap browsing + search filters
  3. Browse menus and reviewsProduct and retailer details
  4. Save favoritesRetailers, products + strains

The person
behind the projects.

Samuel Rombs.

I build tools for working with code and finding answers in technical documents.

srombs@gmail.com ↗