Eli.Jaramillo

Software Engineer

Eli Jaramillo

On my own time I built a real-time multiplayer game, and I run it for 1,300+ players. Three services and 86,000 lines of code. Built solo.

0+players
0in Discord
0K+lines of code
0+ yrslive & solo

About

The short version.

Eli Jaramillo

I'm a software engineer who likes owning the whole problem, from the architecture down to whether the people using it are actually better off. Over the past five years I've shipped enterprise applications, led cross-functional teams, and built the integrations that hold them together across media and financial services.

At NBCUniversal I support payment systems that move more than $6 billion a year, and I write the Python tooling that turns hours of manual reconciliation into minutes. Before that, at Capital Group, I ran a $2.5 million platform build as product manager, scrum master, and QA lead for a nine-person team, and it earned $1.3 million in follow-on funding.

Then there's the project I do for myself. I designed, built, and operate Siliconian Showdown, a live multiplayer game with its own real-time distributed backend. It's where I go deep on the parts of engineering I like most: concurrency, crash recovery, and keeping game state consistent for more than a thousand players at once.

Core stack

TypeScript · Python · React / Next.js · Node / Express · MongoDB · .NET · Socket.IO · SQL

Works in English & Spanish

Flagship project

Siliconian Showdown

A full-stack real-time multiplayer game where you collect, customize, and battle digital creatures in timed turn-based rounds. I started it in 2020 and still run it today, and it's grown from 200 players to more than 1,300. These days I lead a small team on updates, but the architecture and the 86,000 lines underneath it were mine to build.

Siliconian Showdown logo

By the numbers

What it adds up to

1,300+
Registered players
grew from 200, all organic
600
Discord members
active player community
86K+
Lines of code
TypeScript, JavaScript, Python
3
Independent services
client, relay, engine
50+
REST endpoints
plus a custom WebSocket RPC
120+
Playable characters
across 3 warring factions
22
AI configurations
tiers, bosses, and raids
5+ yrs
Live in production
2020 to today
PvP1v1, 3P FFA, 4P FFA, 2v2, casual and rankedPvE Incursionsolo, co-op, 3v1 raidSurvivalendless run, escalating difficultyConquest3-faction, 7-day territory warCustom Roomsplayer-hosted, configurable matches

The centerpiece

How one battle action flows

Anatomy of a turn

Follow one real battle action through all three tiers.

Client

Next.js / React

Everything the player sees and does. It never talks to the engine directly; it only renders state.

Communication Server

Node.js · Express · Socket.IO

All real-time messaging and match orchestration: matchmaking, rooms, spectating, and reconnection.

Game Engine

Python · WebSocket RPC

All game logic: damage, status effects, win conditions, and AI decisions.

trace

$ press “Send a battle action” to trace a turn…

The path is real; the timings are representative. Node owns real-time, Python owns the rules, the client only renders.

Strongest systems

Where the hard problems live

  • Funds are escrowed at match start and settled atomically at resolution using MongoDB multi-document transactions.
  • Settlement is normalized so concurrent payouts, refunds, and rake never race into a double-spend or a lost balance.
  • The invariant holds even when many matches settle at once: total currency in always equals total currency out.

the same invariant payments systems live by

  • If the process dies partway through dispatch, finalize resumes from the last checkpoint instead of paying anyone twice.
  • Rewards across many players and documents are dispatched exactly once, even under partial failure.
  • It was built as a self-contained subsystem on top of the existing PvP infrastructure, added without disrupting live matches.

idempotent, checkpoint-based finalize

  • CombatIQ scores the board against authored rule-trees to pick a move, so every decision is traceable and reproducible.
  • Difficulty tiers and boss personalities are data, not code, which means balance changes ship without a redeploy.
  • It's a deliberate trade-off: ML would look flashier but be far harder to debug, tune, and trust in a live competitive game.

data and rule-trees, not ML

War stories

Bugs that taught me something

Tap a card for the diagnosis & fix.

Experience

The track record.

Most recent first.

  1. Senior Application Engineer

    Dec 2024 – May 2026

    NBCUniversal · Los Angeles, CA

    • Backed enterprise payment systems handling $6B+ in annual revenue across 200M+ transactions, acting as the technical bridge between payment vendors, POS platforms, and internal teams.
    • Led the rollout of 70+ mobile payment terminals across 5 venues, owning device configuration, Oracle POS integration, operator training, and hypercare.
    • Built Python automation for transaction reporting, compliance auditing, and reconciliation that cut multi-hour processes down to minutes.
    • Found and closed a PCI compliance gap where full card numbers were collected over email, working with Infosec and Finance to redesign the process and purge the exposed data.
  2. Consultant

    Dec 2023 – Present

    California Thrift Store · Remote

    Leading the digital transformation of store operations, moving inventory into multichannel retail to lift turnover.

  3. Solutions Engineer II

    Aug 2021 – Dec 2023

    Capital Group · Irvine, CA

    • Acted as product manager, scrum master, and QA lead for a $2.5M web app (.NET/Angular/SQL) consolidating HRIS data for 400+ associates at a $2.5T AUM firm, delivered on time and earning $1.3M in expansion funding.
    • Led a 9-person Agile team through planning, backlog refinement, and biweekly stakeholder demos, shipping features that saved $750K+ a year in manual work.
    • Ran vendor evaluation and selection for a $1M enterprise review platform, from gap analysis through launch in 10 months at 80%+ user satisfaction.
    • Shipped a self-service module that let investment professionals validate their own compensation data, saving operations 1,000+ hours ($150K+) a year.
  4. Post-Closing Intern

    Jul 2020 – Aug 2020

    CoreVest Finance

    Kept loan documentation accurate and consistent to support audit readiness and operational integrity.

  5. Commercial Lending Intern

    May 2019 – Sep 2019

    Pacific Enterprise Bank

    Ran property site visits and financial analyses to validate loan eligibility and sharpen lending decisions.

  6. Research Protections Assistant

    Nov 2017 – Jun 2018

    UCI Office of Research

    Managed research records and compliance documentation so they stayed organized and audit-ready.

EducationBS Informatics · BA Business Economics · University of California, Irvine · 2021

Also built

Contact

Let's talk.

I'm open to engineering and product roles, and to people who like hard problems. Email is the fastest way to reach me.