Brandon Mrgich, profile photo

Brandon Mrgich

Launch Control Systems (LCS) Software Engineer · Amentum at NASA Kennedy Space Center

Software engineer on human-rated launch control systems at NASA Kennedy Space Center — Java GUI platforms, deployment automation, and testing for safety-critical software. Off the clock, a production music catalog and streaming platform.

Skills

Languages

  • Java
  • Python
  • TypeScript
  • JavaScript
  • C/C++

Frontend & UI

  • React
  • Next.js
  • Tailwind
  • JavaFX

Backend, Systems & Infra

  • Node.js
  • Fastify
  • Spring Boot
  • Linux
  • Bash
  • Docker
  • AWS (EC2, S3)
  • Apache ActiveMQ

Automation, CI/CD & Testing

  • Ansible
  • Jenkins
  • Git
  • JUnit
  • Pytest
  • Robot Framework

Data & APIs

  • SQL
  • SQLite
  • MariaDB
  • MongoDB
  • Postgres
  • REST

AI & Agentic Tooling

  • LLM agent orchestration
  • Multi-agent workflows
  • Claude Code / Claude API
  • Prompt & context engineering
  • AI-assisted development pipelines

Selected Work

Most of this runs in private repositories. What follows is the shape of the work rather than a link to the source.

  1. Music Platformprivate source

    Production music streaming, catalog management, and artist services platform.

    6 applications · shared TS packages · edge services · ~199k LOC

    • Architected a Turborepo/pnpm monorepo of 6 applications, shared TypeScript packages, and edge services.
    • Built a centralized admin platform for end-to-end content management — audio assets, releases, metadata, artwork, artist profiles, and rights information behind one interface.
    • Implemented a DDEX-aligned PostgreSQL model in Prisma covering recordings, releases, contributors, royalty ownership, and territorial rights, with integrity enforced at the database level.
    • Developed a secure media pipeline: multipart S3 uploads, content-addressed deduplication, audio metadata extraction, image processing, and Cloudflare edge streaming via signed, range-aware URLs.
    • Built custom auth supporting OAuth, email verification, creator onboarding, and role-based access control across admin and end-user domains.
    • Automated production infrastructure with Docker, GitHub Actions, self-hosted CI runners, EC2, and Nginx for reproducible deployments and centralized secrets.
    • TypeScript
    • Fastify
    • Next.js
    • Postgres
    • Prisma
    • Turborepo
    • Docker
    • AWS
    • Cloudflare
  2. bam-claudeprivate source

    AI agent orchestration and developer tooling system.

    • Engineered a version-controlled configuration and orchestration layer extending an agentic coding system with custom multi-agent workflows, domain-specific skills, and a structured knowledge base — used daily to drive production software delivery.
    • Built a deterministic orchestration engine that decomposes engineering plans into discrete tasks and dispatches them to specialized sub-agents (implementation, testing, documentation, independent compliance auditing), persisting execution state for resumable, auditable runs across sessions and machines.
    • Designed a knowledge pipeline promoting ideas through design essays, executable plans, and living documentation, connected by typed anchor links into a navigable dependency graph — preserving design rationale end to end while routing each agent only the context its task requires.
    • Go
    • Claude Code
    • Multi-agent orchestration
  3. Ephemeral CI Runnersprivate source

    Self-hosted GitHub Actions runners — one disposable container per job.

    • Webhook-driven controller that launches one runner container per CI job and tears it down on completion, managed as systemd units with configurable per-repo concurrency.
    • Runs CI for the music platform monorepo on self-hosted hardware, with a shared pnpm store bind-mounted into every container to keep installs fast across otherwise disposable environments.
    • Python
    • Docker
    • containerd
    • systemd
    • GitHub Apps
  4. Unity Audio Engineprivate source

    Configuration-driven music and SFX engine for Unity 2D — playback behavior authored as data, not code.

    11 engine scripts · game-agnostic core · 3 example adapters

    Built to deliver my own score for Blackthornprod's Pass The Game Challenge — a jam where one game is handed between developers — then extracted from that game as a standalone engine. Three themes and dozens of SFX; the composing credit is on the music portfolio, this entry is the engine that plays it.

    • Modeled playback as four kinds — beds with intensity-driven vertical layering, cues that take over the bed and hand it back, one-shot stingers on either a newest-takes-precedence Event bus or a freely-layered Ambience bus, and state-gated loops — each authored as a ScriptableObject, so scoring decisions are configuration rather than new code per cue.
    • Scheduled every stem through PlayScheduled so layers stay sample-locked to the bed as they come and go. Drift between stems is what makes vertical layering audibly fall apart, and it is not fixable after the fact by fading.
    • Kept the runtime deliberately dumb: mixing is baked in at bounce time, so the engine is a playback system rather than a mixing console. That constraint is what kept it small enough to lift out of the game intact.
    • Separated the game-agnostic core from the game-side drivers that feed it, so the engine never references a game concept. The jam's own intensity, movement, and surface adapters ship as reference examples instead of as part of the engine — which is what made the handoff to the next developer a clean one.
    • C#
    • Unity 6
    • ScriptableObjects
    • Logic Pro
  5. BrainBeatsV4

    EEG-to-MIDI web application — record music with your brain.

    UCF senior design project

    • Led a TypeScript migration of an inherited codebase with a team of four.
    • Redesigned data ingestion to expand device compatibility from two to over a dozen EEG devices.
    • TypeScript
    • React
    • Node.js

Public Repositories

  • dotfiles

    GNU Stow-managed dotfiles — zsh, tmux, git, and starship config kept in one tracked tree.

    Shellupdated Sep 2026

  • nvim

    Neovim config built on lazy.nvim and Mason, with LSP servers and formatters wired up for Java, Python, Lua, and web files.

    Lua★ 1updated Sep 2026

  • VST-prepper

    CLI that turns a raw piano sampling session into a usable instrument: detects transients, identifies the note, normalizes, and writes one file per key across an input tree. Built to cut 88 keys × 4 mic positions without doing it by hand.

    Pythonupdated Aug 2026

Contributed to

  • showell/claude-collab

    Contributed to a multi-agent collaboration tool for agentic coding workflows.

    Go★ 1

Experience

  1. Launch Control Systems (LCS) Software EngineerAmentum · NASA Kennedy Space Center

    · Cape Canaveral, FL

    • Architected and implemented a configurable Java-based GUI platform and application framework that enables independent development, packaging, deployment, and runtime integration of mission applications within a unified operator environment; now the foundation for a growing suite of applications built across the team.
    • Defined framework APIs, application lifecycle standards, shared services, reusable widgets, and common design patterns that standardized development practices across our team and accelerated delivery of operator-facing applications.
    • Designed and developed mission applications and backend services supporting real-time operational issue tracking, telemetry visualization, alarm monitoring, and system status awareness for operators.
    • Led deployment architecture and automation efforts using Python, Bash, and Ansible, creating the first fully automated deployment solution for the GUI software stack after the system previously lacked a repeatable deployment process.
    • Developed a one-button deployment framework capable of provisioning, configuring, and deploying applications, libraries, service definitions, and environment-specific configurations across multiple development and operational Linux environments.
    • Executed dozens of deployments at scale, eliminating manual deployment activity and reducing configuration drift between environments.
    • Modernized legacy operator displays through collaboration with operators and human factors engineers, improving information presentation and enabling personnel to identify critical events and system state changes with substantially reduced latency and information overload.
    • Established automated GUI testing infrastructure and expanded unit, integration, and functional testing capabilities through JUnit and Robot Framework, increasing validation coverage for safety-critical software releases.
    • Performed JVM profiling and performance optimization using VisualVM and custom analysis techniques, identifying memory leaks and high-footprint applications while implementing lazy loading, deferred initialization, and paging strategies that reduced memory utilization by approximately 80%.
    • Standardized performance optimization techniques through shared framework libraries, enabling other developers to adopt memory-efficient patterns without duplicating implementation effort.
    • Mentored nearly a dozen interns and junior engineers through architecture discussions, design reviews, implementation guidance, testing practices, and software engineering standards while leading initiatives focused on framework standardization and deployments.
  2. Launch Control Systems (LCS) Software EngineerJacobs Space Operations Group · NASA Kennedy Space Center

    · Cape Canaveral, FL

    Same Launch Control Systems role and program — the support contract transitioned to Amentum in October 2024.

  3. Spaceport Command and Control Software Development InternNational Aeronautics and Space Administration (NASA)

    · Cape Canaveral, FL

    • Contributed to a Class A, human-rated, safety-critical command and control system supporting real-time launch operations.
    • Designed and implemented automated black-box tests integrated into CI/CD pipelines, reducing end-to-end test execution time from days to hours and enabling faster validation cycles.
  4. Application Development and Maintenance InternCognizant

    · Remote US/Canada

    • Developed enterprise software using Java Spring Boot, Maven, and Gradle.
    • Developed Node.js applications deployed on AWS.
    • Deployed and maintained Node.js and Spring Boot services.

Education & Recognition

  • University of Central Florida

    B.S. Computer Science

    Jun 2017 — Dec 2023

  • NASA OSTEM: Shining Star Award nomination

    Recognized for leadership, work ethic, and technical expertise.

    Aug 2023