red.systems / home

~ currently in Manila, PH

Red Gabriel
Tagura

Computer Science — Systems Programming & Backend Engineering

I build the things beneath the things. Memory allocators, compilers, custom servers, and the backend scaffolding that makes software not embarrassing to run. Most of my work lives in Zig, C++, or Python — whichever one the problem actually calls for.

Currently a CS student at PUP Manila (GWA 1.35, President's Lister), finishing an undergraduate thesis on profile-guided memory allocation. On the side, I head the backend department at Shift Out and run membership infrastructure for Junior Blockchain of the Philippines.

I believe fast, honest tools are also good tools. The simplest thing that could possibly work at scale.

Browse to

01

Technical Skills

Languages

Zig C++ C Python JavaScript Java PHP

Frameworks & Libraries

Express.js FastAPI Node.js React Native N-API Raylib Java Swing Laravel

Cloud & Infra

Docker Git CI/CD Vercel Render Microsoft Azure

Databases

PostgreSQL Supabase Redis MySQL
02

Projects

Zig Python Node.js · 2026 – Present

Stratified Autonomous Memory Manager

Undergraduate thesis. Offline profile-guided allocator using mmap with K-Means lifetime prediction to pre-bake arena selection — targeting P99 latency and peak RSS reduction in 1 GB-constrained Node.js backends.

SystemsMemoryThesis
→
Zig Raylib · 2026

ambigui2 — Tetromino AI Engine

Dual-state bitboard Tetris engine in Zig with Raylib rendering. 139k nodes/sec at depth 5, 17.4ms/move. Heuristic weights evolved via tournament genetic algorithm, improving score 2.70× over baseline.

AIGame EngineAlgorithms
→
Zig TCP Linux · 2026

Zig Server — Custom HTTP/1.1 Web Server

Zero-dependency HTTP/1.1 server in Zig. Raw TCP socket listener, hand-rolled HTTP parser, Linux epoll for non-blocking async I/O. No libuv, no framework — the simplest thing that works at scale.

SystemsNetworking
→
React React Native Supabase Llama-3 · 2025

Gabay Med — AI Healthcare Platform

Two-sided platform: hospital middleware dashboard for insurance and appointments, plus a patient mobile app. Dual-Agent RAG pipeline on Llama-3 with triage and DOH-cross-validation agents.

AIHealthcareFull-stack 🏆 2nd Runner-up — PUP Hackathon
→
C++ Python N-API · 2025

r-alloc — Custom Memory Allocator

Arena and Slab strategies in C++ via mmap, bridged to Node.js through N-API to bypass V8 GC. Reduces maximum latency and peak RSS vs glibc malloc and V8.

SystemsMemoryC++
→
C++ Compiler Design · 2025

Setra — Domain-Specific Language

Full compiler pipeline in C++ from scratch — Lexer, Parser, Semantic Analyzer — for a DSL targeting discrete mathematics. No frameworks, no generators.

CompilersLanguages
→
View all projects →
03

Experience

Shift Out Sep 2025 – Present

Backend Department Head · Manila, PH

  • Designed and delivered Behind the Code workshop series covering backend architecture, Git branching, and CI/CD pipelines — directly upskilling 50+ members across 4 cohorts.
  • Established organization-wide code review standards and PR workflow policies, reducing merge conflicts and integration defects across all active tech projects.
Junior Blockchain of the Philippines Sep 2025 – Present

Membership Committee Head · Manila, PH

  • Architected a centralized membership platform on Supabase (PostgreSQL), consolidating onboarding, attendance, and engagement tracking for 400+ active members into a single auditable data source.
  • Designed row-level security policies and relational schemas to enforce role-based access control across all membership lifecycle events.
04

People Also Ask

Mostly low-level systems work — memory allocators, custom servers, compiler pipelines. Right now the thesis is the main project: a profile-guided allocator in Zig that tries to predict allocation lifetimes using K-Means so it can pre-assign arenas before runtime. On top of that, org work at Shift Out and JBP.
Zig gives you C-level control without the footguns of C, and without Rust's sometimes-oppressive borrow checker when you're doing intentionally unsafe memory work. The comptime system is genuinely interesting. That said, I'll use whatever makes the most sense — the allocator uses Zig, the ML side uses Python, the bridge uses N-API. Tools should serve the problem.
Yes. Particularly interested in roles involving infrastructure, developer tooling, backend architecture, or anything close to the metal. Reach out at redtagura07@gmail.com.
Third year BS Computer Science at Polytechnic University of the Philippines — Manila. GWA 1.35, President's Lister. Expected graduation June 2027. The thesis is the current capstone work.
An AI-powered healthcare platform built at the PUP Hackathon. Hospital-side: a dashboard for managing doctors, appointments, and insurance. Patient-side: a mobile app for booking. The interesting part is the dual-agent RAG system on Llama-3 — one agent routes the patient to the nearest qualified hospital, another cross-validates recommendations against DOH guidelines. Took 2nd Runner-up out of 18 teams.
Yes — the /blog section has deep-dives on the technical decisions behind each project. Each project page also links to the corresponding write-up.
05

Links

[ email ] Email redtagura07@gmail.com [ github ] GitHub github.com/Xuinnz [ linkedin ] LinkedIn red-gabriel-tagura [ phone ] Phone +63 994 426 2357

Currently based in Sta. Mesa, Manila · Open to remote & on-site opportunities