Felipe Oliveira

Unity Gameplay Programmer.

I build gameplay systems, tools, and in-game UI in Unity, with a focus on performance.

View experience
  • Gameplay Systems
  • C#
  • Performance
  • Networking & Serialization
  • Game UI

Experience

Montage of Dead Drop gameplay scenes from Blue Gravity.
Nov 2025 – Present

Unity Developer

BGNS Studios · Full-time

Remote

Dead Drop

Coming soon
  • Steam
  • Single-player
  • Action roguelike
  • Procedural map
AI · reference, autocomplete & agents

From the beginning of the project, I frequently used AI for reference and consultation, alongside GitHub Copilot autocomplete. In recent months of development, there has been greater encouragement to use AI agents for programming and game development.

Dead Drop is a single-player action roguelike in development at Blue Gravity. It combines combat, abilities, and a procedurally generated map, offering different environment layouts with each run.

I have been involved since development began, when the programming team consisted of one other developer and me. As the team grew, I continued contributing to core gameplay systems, optimization, tools, and environment creation.

1 / 6 · City overview and rooftop combat

Effect System

Featured contribution · gameplay architecture

A shared foundation for weapons, items, and abilities.

I developed a modular effects system inspired by Unreal’s Gameplay Ability System, focused on attributes and gameplay rules. Built during the first months, it has been used throughout the project for weapons, items, and abilities, from simple damage to complex interactions.

Attributes & tags

Modifiers for health, shields, experience, and currency, with tag filters representing factions, ability types, and other contexts. Modules can also grant and modify tags.

Composable rules

Stacking, duration, and periodic applications combine with conditions that activate sub-effects, building complex behaviors from modules.

Player feedback

Effects communicate gameplay events through particles and sounds: fire can add a particle effect, while a damage event can trigger a specific sound.

Diagnostics & maintenance

Logs help identify what happens as effects are applied. The modular structure combines responsibilities for different gameplay needs.

One rule, multiple conditions

Fire ability + target from a specific faction → bonus damage. Another condition can activate a sub-effect while health is below a threshold. These examples show how tags and attributes express contextual rules.

Performance by design: I designed the core and many modules to operate without runtime memory allocations, minimizing garbage generation. Built during the first months, the system became a foundation of gameplay throughout development.

In tests I monitored during development, the core logic processed more than 8,000 damage applications in a single frame without exceeding 1 ms. This measurement refers to the system’s core, not total frame time.

Enemies, projectiles & optimization

I developed enemy navigation, movement, and attack systems, as well as the projectile system. I also contributed to performance improvements across different areas of the game.

Tools & shaders

I created shaders and development tools to support game designers, system configuration, and game balancing.

3D modeling & environment composition

I modeled streets, sidewalks, building blockouts, bridges, structures, and platforms. For the platforms, my contribution was modeling, without texturing.

I was responsible for assembling the chunks used in the procedural map: street layouts, terrain preparation and painting, and placement of props, rocks, and effects. This work brings together assets from different artists to build the game’s environments.

Lyra’s Revenge

Contribution at Blue Gravity · Mountainroad Games

  • PC
  • Unity
  • Rhythm FPS
  • Shaders
  • Post-processing
  • Level creation
Game website ↗

A rhythm FPS by Mountainroad Games where the player builds the soundtrack while fighting. I contributed briefly to the project at Blue Gravity, implementing a few small gameplay mechanics, with most of my work focused on the game’s visuals.

  • Post-processing

    Tuned the post-processing and configured tone mapping, color grading, ambient occlusion and bloom to refine the game’s look.

  • Environment shaders and effects

    Created environment shaders and visual effects, including a laser grid marking the dead zone. I also programmed custom post-processing effects for the game and effects for changes in the environment’s colors.

  • Animated skybox

    Configured an animated, pixelated skybox to resemble a giant screen, reinforcing the feeling of being inside a simulation.

  • Level creation

    I also contributed to creating levels for the game, working with the team.

SIS-ASTROS room with the simulator displayed on large screens and an interactive tabletop.
Jul 2023 – Jul 2025

Unity Developer

SIS-ASTROS · Full-time

On-site in Santa Maria, Rio Grande do Sul, Brazil. Technologies: C# and Unity.

SIS-ASTROS GMF is a multiplayer training simulator developed by the Federal University of Santa Maria (UFSM) in partnership with the Brazilian Army. It integrates three connected platforms and includes a touchscreen tabletop for collaborative map exploration. Access is not public: it is restricted to the Armed Forces and Avibras. The simulator is used for tactical instruction related to Avibras products.

AI · Reference only

During my work on SIS-ASTROS, I used ChatGPT only as a reference. I did not use AI agents to write code.

  • Technical decisions · Unity
  • Multiplayer
  • Network migration
  • Optimization
  • UI
  • Tooling
  • Gameplay

ICEIS 2025

Enhancing Usability in Large Map Interactions: A Novel Magnifying Lenses Technique for Tabletops

DOI 10.5220/0013300100003929

While working on SIS-ASTROS, I helped develop the tool studied in this paper, which I co-authored. The tool uses a digital magnifier to enlarge map details on the interactive tabletop while keeping the surrounding area visible.

View contributions
  • Technical decisions · Unity

    Served as a Unity technical resource for the team, helping assess feature feasibility and guide implementation decisions through my knowledge of the engine.

  • Multiplayer

    Worked on the legacy UNet/HLAPI networking layer, with responsibilities distributed across three platforms. Built a source generator to automate network commands, serialization, and data conversions on send and receive, reducing boilerplate and enabling authority to be assigned to different stations, including non-host stations.

  • Network migration

    Actively contributed to technical decisions and refactoring for the migration from legacy UNet/HLAPI networking to Netcode for GameObjects, helping the team adapt the system. My involvement continued until I left the project, before the migration was completed.

  • Optimization

    Contributed to performance improvements across different areas of a complex simulator, helping optimize its runtime behavior.

  • UI

    Developed menus and UI features to help users interact with the simulator’s extensive functionality.

  • Tooling

    Created development tools to support the team’s workflow and simplify recurring project tasks.

  • Gameplay

    Contributed to simulator mechanics and features, and investigated and fixed bugs across the system.

UI · menu tabs proof of concept

I developed this proof of concept to explore tabbed menus in SIS-ASTROS. The feature was subsequently implemented and used in the project. Menu navigation is central to the simulator, providing access to a large share of its functionality.

Internal workshop · Unity interfaces

As part of the team’s knowledge-sharing workshops, I led a session on building interfaces in Unity. I prepared this sample interface to demonstrate how it was built and explain layout, pivots, anchoring, and the organization of UI elements.

Sample interface prepared for the internal workshop, with panels and layout guides in Unity.

Projects

Technical challenge · Unity

A game in 48 hours

AI · Autocomplete

For the 48-hour project, AI assistance was limited to code autocomplete. I did not use AI coding agents.

I built this Unity prototype as a technical assessment during the hiring process that led me to Blue Gravity Game Studio, where I currently work.

  • Slot-based inventory
  • Item interaction and pickups in the game world
  • Save and load system

The assessment allowed the use of assets. I used assets for characters, animations, environments, icons, and UI, plus a texture pack for the borders.

View code on GitHub
Original build screenshot showing the inventory, equipment, and options panels.

Screenshot from the version submitted for the 48-hour technical assessment.

GameplayTags — tag creation, hierarchy, and selection in the Unity editor.

Community interest

Stars and forks reflect the community’s interest in following, exploring, and adapting the project.

GitHub data · Last checked: 11/13/26

Unity Package · BandoWare

GameplayTags

A Gameplay Tags implementation for Unity, inspired by Unreal Engine’s system. It recreates features such as hierarchical tags, containers, queries, and tag counts with callbacks.

Learn more about Gameplay Tags

Gameplay Tags are labels that describe what is happening in a game. A character can have several tags at once. For example, State.Stunned and State.Silenced belong to the State family, which can be queried as a group.

How does it work in play?

When a hit applies State.Stunned, the combat system can check that tag to block attacks. The UI can listen for changes to the same tag’s count to display a stun icon. If two effects apply the tag, the icon can stay visible until the last effect ends.

How does this help a project?

The package lets you combine conditions and reuse the same rule across abilities, enemies, and effects. A new trap, for example, can stun the player by applying the same tag without requiring changes to the UI or the code that blocks attacks.

AI · Autocomplete

AI assistance on this project was limited to code autocomplete. I did not use AI coding agents.

  • Unity
  • C#
  • Source Generator
  • Roslyn
  • Tags available directly in code

    The source generator reads assembly attributes to generate hierarchical C# classes, providing strongly typed tag access and reducing boilerplate.

  • A foundation for gameplay systems

    Tags represent states, abilities, and effects and help express gameplay conditions. They are useful in systems such as Gameplay Abilities, an area I am interested in exploring.

I have been dedicating less time to maintaining this project recently.

View code on GitHub

Academic project · 2022

Subida das Pilhas 2022

No AI used

I did not use AI tools in the development of this work.

  • Android
  • Mobile
  • Unity
  • 2D
  • Academic project

A 2D mobile arcade game built with Unity and released for Android in 2022. The player controls a small planet that climbs platforms, collects batteries and takes them to the correct disposal points. Jumps are automatic, while horizontal movement is controlled by touch. The mechanics encourage responsible battery disposal.

Developed as a team project in a supplementary course offered by the game development program, which I took during my Computer Science degree.

My contribution

I was responsible for most of the programming and all of the game’s user interface (UI).

Play and download the APK on itch.io

Personal project

GTA Vice City Files in Unity

No AI used

I did not use AI tools in the development of this work.

A panoramic view of GTA Vice City cars, boats, planes, and helicopters lined up in Unity.
A showcase of 107 GTA Vice City vehicles imported from the original files and arranged in Unity.

I built a C# pipeline that uses the gta3.dir index to extract 3D models (DFF) and textures (TXD) from gta3.img and import them into Unity. I used default.ide to match each vehicle to its model and textures and generate reusable prefabs. I enjoyed deserializing and interpreting these files to see how a game of this scale structures its data.

I integrated the original day-night cycle data to recreate the game’s sky colors, ambient lighting, and fog in Unity. I also recreated the reflection effects on the vehicle bodies.

Game Jams

Going Shopping with Josué Veleda — gameplay

14ª Game Jam Santa Maria · 2023

Going Shopping with Josué Veleda

  • In person
  • 48h
  • Theme: shopping
No AI used

No AI was used in the development of this project.

A top-down action game where supermarket purchases become weapons against hordes of demons. Created with friends from SIS-ASTROS during the 14th Game Jam Santa Maria, September 29–October 1, 2023.

I used this in-person jam to practice 3D: I created roughly 90% of the 3D art, including characters and environments. Other team members made some items, and I also helped with programming and implementation.

Team: Adrian Kaminski · Felipe Marques · Gabriel Schuster · Gustavo M. de Freitas · Josué Veleda

🎸 Soundtrack by Adrian Kaminski, our team’s guitarist!

Non Linear Menace — gameplay

12ª Game Jam Santa Maria

Non Linear Menace

  • Solo development
  • Theme: non-linear
  • Top-down
No AI used

No AI was used in the development of this project.

I developed this top-down survival game solo for the 12th Game Jam Santa Maria. The player controls a tank whose shot follows a random path, exploring the non-linear theme.

The trajectory is previewed visually before firing. Each shot generates a new path: the challenge is to position the tank to use that trajectory and hit the targets.

Participant: Felipe Marques de Carvalho de Oliveira.

Talks and workshops

Guest instructor

Game development workshop

16th Computing Symposium (SIRC) · Universidade Franciscana (UFN)

At the invitation of Professor Fabrício Toledo Londres, I led an introductory game development workshop. I built a small game to demonstrate foundational concepts in practice.

View the event on UFN’s website
Stories from the workshop

Education

Bachelor’s degree in Computer Science

UFN

Undergraduate thesis

CrowdMorph — Crowd animation in Unity

No AI used

I did not use AI tools in the development of this work.

  • Animator feature conversion

    The system converted Unity animation data, including state machines, parameters, transitions, Avatar Masks, and override and additive layers. Blend shapes and Humanoid animations were not supported; Humanoid support remained future work.

  • Unity DOTS / ECS

    Converted Unity animation components and assets to a data-oriented model for batched, parallel processing.

  • Compute shaders (GPGPU)

    The CPU updates animation states and sends commands to the GPU, which computes skeletal transforms in parallel.

  • Linear Blend Skinning

    The GPU deforms each character mesh by blending bone transforms according to each bone’s weight at each vertex.

Visual comparison

One entity animated by CrowdMorph and one GameObject using Unity’s Animator. In this comparison, the visual results of the two systems are virtually indistinguishable.

In the thesis benchmark with 1,600 characters, CrowdMorph averaged 200 FPS versus 9 FPS for Unity’s built-in module, with the same visual result in the tested scenes.

Performance comparison

CrowdMorphUnity’s built-in module

Average FPS · higher is better

Performance comparisonAverage FPS comparison. At 1,600 characters: CrowdMorph, 200 FPS; Unity, 9 FPS.0501001502002502550100200400800160025 characters · CrowdMorph: ≈ 230 FPS50 characters · CrowdMorph: ≈ 230 FPS100 characters · CrowdMorph: 225 FPS200 characters · CrowdMorph: 222 FPS400 characters · CrowdMorph: 207 FPS800 characters · CrowdMorph: 202 FPS1600 characters · CrowdMorph: 200 FPS25 characters · Unity: ≈ 225 FPS50 characters · Unity: ≈ 200 FPS100 characters · Unity: 130 FPS200 characters · Unity: 85 FPS400 characters · Unity: 59 FPS800 characters · Unity: 20 FPS1600 characters · Unity: 9 FPS

Animated characters

Source: thesis · Figure 4

The performance gain comes with a trade-off: bone transform calculations and skinning were moved to the GPU, leaving the bone hierarchy unavailable on the CPU. This enables large crowds with high visual fidelity, but limits systems that need CPU access to bones.

2016 – 2022
Video in Portuguese, recorded for my professor. Gameplay mechanics, Blueprints, and materials/shaders.
Watch on YouTube ↗

Academic project · Unreal Engine

Sword — Unreal Engine project

Engine and development

  • Unreal Engine
  • Blueprints
  • Materials / Shaders

My main academic project with Unreal Engine: a simple game where you can control platforms using Q and E. Recorded for my professor, the demonstration shows some of the mechanics and the visual graphs behind the gameplay logic (Blueprints) and materials and shaders. The video is in Portuguese.

Although my career has focused on Unity, I also have a strong interest in Unreal and C++. I have spent considerable time exploring the engine’s source code to learn from its patterns and implementation choices, particularly in memory management, cross-platform support, and optimization.

Animated preview of the Amamenta blog and its posts.
Website preview

Academic collaboration · UFN

Amamenta

Developed the blog and admin panel for Amamenta, a project associated with UFN’s Master’s program in Maternal and Child Health. The platform lets the team publish and manage breastfeeding guidance articles and administer application resources.

Front-end

  • React
  • Sass
  • Semantic UI

Back-end

  • NestJS

Database / ORM

  • MySQL
  • Sequelize
View code on GitHub
No AI used

I did not use AI tools in the development of this work.

3D art & textures

A collection exploring my interest in environment modeling and hand-painted textures. None of the work shown in this section was AI-generated.

Contact

Let’s talk

Get in touch by email to discuss projects, collaborations, and opportunities.