KALEB KIRCHER

Mobile platform for police, fire, and EMS first responders

Principal Engineer and mobile lead for Android and iOS at Oracle Public Safety, building the field-facing apps that police, fire, and EMS agencies use for in-vehicle CAD dispatch, records management, and real-time incident response. Modernizing the Android stack from legacy Views to Jetpack Compose and restructuring the codebase into a modular monolith with KMP modules targeting Android, iOS, and Windows from shared Kotlin. Refactored the OAuth2 IDM SDK, rebuilt the WebSocket layer for reliability and background compliance, and migrated the metrics SDK to offline-first architecture. Designed the CI/CD pipeline with shift-left testing and automated release trains. Leading the integration of agentic AI engineering — subagent teams, orchestrated pipelines, code maps, quality ratchets, live simulators, and automated testing tools.

In-cab guidance and field management for precision agriculture

Lead engineer and eventual architect on Trimble's flagship in-field guidance platform, shipping on GFX Android-based displays across mixed-fleet farm operations worldwide with 300,000+ daily active users. Built the OpenGL ES 3D rendering engine powering the Run Screen — the real-time guidance view operators use 12+ hours a day — and rewrote it for a 200x performance gain. Designed and implemented FreeForm (custom guidance patterns), AutoSync (automatic cross-device data sharing), and Square Headlands (corner coverage optimization). Led the teams that shipped NextSwath end-of-row turn automation. Took crash-free rates from 52% to over 99%. Owned the full UI layer, overhauled the SQLite databases, and rebuilt the import/export file pipeline. Designed an ETL pipeline with AWS (SQS, Glue, Flink, SageMaker) for diagnostic telemetry at scale.

Automotive ECU calibration and optimization suite

Compose Desktop application for calibrating Bosch engine control units across four platforms: ME7, MED17, Motronic 3.8x–5.9x, and MED9. Implements a three-stage workflow — Configuration, Calibration, and Optimization — with calculators for fueling, injector scaling, torque/load tables, ignition timing, and boost control. The log-analysis optimizer diagnoses boost control and volumetric efficiency errors from datalogs, then generates corrected calibration maps. Parses TunerPro XDF and WinOLS KP binary formats. Ships standalone on macOS, Windows, and Linux from a single Kotlin/Compose Desktop codebase.

Multiplatform cellular automata simulator and art generator

Kotlin Multiplatform cellular automaton explorer shipping on Android, iOS, and Desktop from a shared codebase. Implements 300+ rules across 14 categories: Wolfram elementary (1D), Life-like (B/S notation), totalistic, continuous (Lenia), neural network-inspired, social behavior models, physics simulations (Wireworld, Ising model, diffusion-limited aggregation, forest fire), probabilistic rules, fractals, and a Turing machine simulator with classic programs like Busy Beaver. Renders in real-time via OpenGL on Android/Desktop and Metal on iOS, with 17 color schemes, 9 lattice topologies (square, hexagonal, triangular, 3D cube), and artistic post-processing effects including bloom and trail decay. Exports stills as PNG and animations as GIF.

Sensor filter and fusion for Android and iOS

Kotlin Multiplatform library providing device-agnostic sensor fusion for Android and iOS. Implements six fusion algorithms — Madgwick (gradient descent, 6-DOF/9-DOF), Mahony (proportional-integral), Extended Kalman Filter, Complementary Filter, Linear Kalman Filter, and Low-Pass Filter — each producing orientation quaternions and gravity-compensated linear acceleration. Includes three smoothing filters (low-pass, mean, median) with configurable time constants, a GPS Kalman Filter for position/velocity/acceleration estimation, and WGS84 coordinate conversion utilities between geodetic, ECEF, and local ENU frames. Published to Maven Central and Swift Package Manager.

Android emulator and device lease broker

Lightweight CLI tool written in Go that manages a shared pool of Android emulators and physical devices for local development, agentic coding sessions, and CI runners. Jobs lease a device with `roboranch with-lease`, which exports ANDROID_SERIAL, runs the command, then cleans and releases the emulator automatically. The cleanup policy uninstalls third-party packages, kills processes, clears logcat, and resets animation scales. Supports self-hosted GitHub Actions runners with a persistent warm pool across jobs, and job-local configs for hosted runners. Distributed via Homebrew tap and GitHub Releases.

Repository mapping and route packs for coding agents

Python CLI tool that builds readable, navigable maps of codebases and generates task-specific route packs for AI coding agents. Produces a `.skyhook/` directory containing an INDEX.md master catalog, structural maps (map.md, map.json), documentation summaries (docs.md), architecture overviews, test inventories, and focused area guides. Six built-in profiles — product planning, requirements planning, technical breakdown, implementation, code review, and bug hunt — each generate a tailored route pack scoped to the task at hand. Designed to give agents the minimum context they need to work effectively without flooding their context window.

Project tuning and ratchet layer for coding agents

Python CLI tool that builds a persistent project ratchet on top of Skyhook's code maps. Generates onboarding docs capturing product intent and architecture direction, lightweight quality analysis (churn, coupling, volatility), documentation gap inventories, and agent team definitions — all under a review-first `.burnplan/` directory. Proposals live in a staging area until explicitly promoted, so generated documentation never silently becomes project truth. Includes a worklog and rationale ledger that records what agents changed and why, giving every future session the context previous ones lacked. Ships with opinionated Product Owner and Project Manager team profiles, editable via `teams.json`.

Native Android and iOS vehicle diagnostics and performance tracking

Native Android and iOS app for vehicle diagnostics, live telemetry, and performance sessions. Built around a shared clean architecture with platform-native Jetpack Compose and SwiftUI screens, offline-first storage, and cloud sync. Its OBD2 engine manages adapter connections, PID polling, unit conversion, diagnostic trouble codes, VIN lookup, and live gauge streams, while performance tools record acceleration runs and session history.

Native Android and iOS dyno, drag timer, and datalogger

Native Android and iOS performance app that turns a phone and OBD2 stream into a mobile dynamometer, drag timer, and datalogger. Built with shared domain logic and native Jetpack Compose and SwiftUI interfaces. Dyno calculators estimate wheel horsepower and torque from RPM sweeps, gearing, vehicle mass, and correction inputs; drag mode records speed intervals and quarter-mile passes; the datalogger charts live channels and overlays runs for comparison.

Native Android and iOS GPS lap timer with predictive splits

Native Android and iOS track-day app for GPS lap timing, predictive deltas, and group sessions. Built on shared timing logic with native Jetpack Compose and SwiftUI experiences, offline track and session storage, and cloud-backed sharing. Its sensor-fusion timing engine combines GPS, IMU, interpolation, and virtual finish lines to detect laps and splits reliably, while live deltas, leaderboards, track editing, and camera HUD overlays support real sessions on custom or community circuits.