Austin Rife

Engine Programmer

When the box isn't big enough, I build a new one. Preferably in Rust.

Rust C++ HLSL Vulkan Unreal Engine

About Me

I build the systems at the foundation. Engine architecture, core systems, performance-critical code, and the tools that tie it all together. The low-level and broad reaching work where getting it wrong means everything above is wrong too.

Professionally, I've shipped four VR titles across two studios, with a consistent thread of performance optimization and developer tooling. I owned the cinematics systems for three of them, evolving the design across each iteration.

I'm drawn to the hard problems that few want to touch: software architecture, systems design, compilers, graphics pipelines. The foundational work that rewards careful thinking and quietly makes everything else possible.

Portfolio

Unannounced VR Title

AAA VR project under NDA. Deep engine and systems work. Spatial partitioning, audio streaming, engine optimizations, and cinematics system v2.

Contributions

Cinematics System v2 — Ground-up rewrite of the cinematics tooling, building on lessons from v1.
Automated IGC Pipeline — Built fully automated cinematic rendering pipeline capturing 360 video and posting to SyncSketch for review.
Spatial Partitioning — Hierarchical spatial data structures for efficient world queries and streaming.
Audio Streaming — FMOD bank loading and audio memory optimization.
Motion Matching Investigation — Evaluated feasibility of motion matching in Unreal with animation team. Abandoned due to system bugs.
Engine Optimization — Ongoing performance analysis and low-level optimization work.
AI Workflow Integration — Explored integrating AI tools into production workflows.

Context

Unreleased AAA VR title. Unable to share further details due to NDA.

Asgard's Wrath 2

AAA VR action RPG for Meta Quest 2/3. 60+ hour Norse mythology epic. Built cinematics system, save system, AI prototype. Executed major engine upgrade.

Contributions

Cinematics System — Designed and built custom node-graph cinematics tooling from scratch. Scaled team adoption from 3 to 30 users. Shipped a Pixar-movie's worth of cinematic content.
Save System — Sole owner of save/load architecture handling complex world state across 60+ hours of gameplay.
HTN AI Prototype — Wrote the hierarchical task network prototype that evolved into the game's AI systems.
Engine Upgrade (UE 4.25→4.27) — Primary engineer for 6 months of a year-long major engine upgrade, merging extensive custom engine changes and ensuring stability at switchover.
Packaging Optimizations — Custom smart world pointer implementation for reduced package sizes.
Performance — Ongoing performance analysis, optimization, and Blueprint VM profiling work.

Context

Award-winning 60+ hour action RPG for Meta Quest 2/3. One of the largest and most ambitious VR titles to date, pushing the boundaries of standalone hardware.

Asgard's Wrath

AAA VR action RPG for Oculus Rift. Award-winning Norse mythology adventure. IGC systems, NPC logic, paired animations, and gameplay mechanics.

Contributions

IGC Systems — Owned the in-game cinematic systems, building on inherited design from another engineer.
NPC Logic — Implemented fairy companion and tavern NPC behavior systems.
Paired Animations — Built paired animation support used across narrative setpieces throughout the game.
Lock & Key System — Wrote the lock and key mechanism used on all doors in the game.
Buoyancy Integration — Integrated buoyancy physics into opening IGC with PID-based following for convincing water movement. Resolved obscure undefined behavior.
Gameplay Mechanics — Wrote the explosives mechanic for destructible props.
Performance — Ongoing performance analysis and optimization.

Context

Award-winning VR action RPG that set a new standard for depth and polish in VR gaming. Built for Oculus Rift, later updated for Quest 2 via Air Link.

Marvel Powers United VR

Cooperative VR brawler for Oculus Rift. 4-player online co-op with iconic Marvel heroes. Joined late to help ship.

Contributions

Critical Bug Fix — Diagnosed and resolved an extremely obscure threading/networking/GPU sync bug over 1-2 months. Critical path blocker.
Performance — Performance analysis and optimization.
Ship Support — Various bug fixes and data upgrades as needed to get the game out the door.

Context

Cooperative VR brawler letting players embody Marvel superheroes. 4-player online co-op with a roster of iconic characters. Joined the project near completion to help push it across the finish line.

Cyberpong

Futuristic VR pong for HTC Vive. Arcade and home release. Networked AI, localization, and arcade tooling.

Contributions

Networked AI — Rewrote the single-player AI system to support networked multiplayer matches.
Localization System — Built localization infrastructure from scratch (Unity lacked built-in support at the time).
VR Arcade Tooling — Wrote external setup tool for VR arcades and configured game for arcade deployment.

Context

Futuristic VR pong for arcade and home VR. 3-month stint during the early days of consumer VR.

Eden

Personal game engine. 7 years, 75+ crates, built almost entirely solo before AI coding tools existed. Ground-up systems exploration.

Highlights

Custom ECS - Data-oriented entity component system with runtime archetype queries.
Vulkan Renderer - GPU-driven rendering architecture with bindless resources.
Asset System - Thread-safe objects-with-properties system with hot-reloading and incremental compilation.
EFX Shader Language - Metalanguage on top of HLSL providing structure and heavy codegen for shader authoring.
Silk UI Framework - Custom reactive UI framework built from scratch.

Scope

75+ crates covering a custom standard library, math, platform abstractions, job system and threading primitives, serialization (including custom binary format), virtual file system, build tooling, and more.

Why

Understanding engines at a level most developers never want or need to. When you've built the entire stack yourself, debugging someone else's becomes a lot easier. Strong opinions, formed by actually building instead of just using.

Agentic RenderDoc

AI-powered graphics debugging through MCP. Agents inspect draw calls, extract buffer and texture data, query pipeline state, diff GPU events, and navigate the RenderDoc UI conversationally.

Highlights

MCP Integration — Bridges any MCP-capable agent with RenderDoc's replay engine via an MCP server and RenderDoc extension communicating over a local TCP socket. Handles multiple open RenderDoc instances and permits full exploration of a capture.
Three-tool Architecture — Eval, Search API, and Instance. Three semantically-grounded tools replace the original 70-tool design, backed by findings from Zhang et al., "One Tool Is Enough" showing fewer well-documented tools outperform large tool suites for LLM agents.
Battle-tested — A prototype version was used to track down real rendering bugs in Eden, which directly informed the Agentic RenderDoc design alongside the research paper.
AI-Written — Both iterations were written entirely by AI. Human contribution focused on design, planning, testing strategy, optimizing LLM interactions, and driving an iterative refinement process.

Context

An experiment in AI-driven development. The original prototype was a conventional MCP server that exposed 70 tools that mapped 1:1 to RenderDoc's exposed features. Lessons from that real-world use case, combined with research from Zhang et al., drove the Agentic RenderDoc down to three tools with rich descriptions, producing better results with less complexity. Open sourced under the Eden Labs organization.

Contact

Interested in working together? Feel free to reach out.

Email LinkedIn GitHub

Currently exploring opportunities in game engine development, simulation technology, and systems programming.