My name is Martin Mihov

Moonfrost

Platform(s):Windows

Technology:Unity, C#, Nodejs, Javascript

Description:

I joined as a Senior Software Engineer in the early development stages, implementing key game features like character controls, animations, item behaviors, UI management, SFX/VFX systems, asset loading, scene transitions, game modes, and inventory management. I collaborated with the backend engineer on networking features and contributed to the transition to a fully server-authoritative multiplayer system, implementing logic and validation. With a small engineering team, we owned our features end-to-end, refining them based on team feedback. I also worked closely with design, art, and audio teams, creating tools to empower them to configure the game without engineering support.

Forza Street

Platform(s): UWP, Android, iOS

Technology: Unreal Engine 4, C++

Description:

I was a major part of the development team throughout the whole development lifecycle of Forza Street and became the main Gameplay Programmer. Some of my contributions include the game’s AI, interaction model, physics modifications and matchmaking.

The work involved:

• - Working closely with the design team while implementing features and creating tools for them to be able to balance and tweak the gameplay.
• - Creating custom shaders, vfx and tools to support the artists and the art pipeline.
• - Working closely with production and project leads on planning and estimating tasks.
• - Doing daily code reviews and ensuring quality and performance of the game code.
• - Mentoring new employees while they get familiar with the existing code base and work process.

Hot Wheels id

Platform(s): Android, iOS

Technology: Unity, C#

Description:

I worked on the physical play side of Hot Wheels id. Creating and iterating on the interaction model between the physical toys and the mobile game.

Popping Ballons

Platform(s): Android

Technology: Android SDK

Description:

My first game ever.

The goal is to pop the balloons by tapping on them before they manage to reach the top. There are also clouds which prevent you from popping balloons. The difficulty of the game increases slightly with time.

It was developed using the Android SDK and was targeted for Android 2.1. It also had a leaderboard and achievments, which are currently not working due to the service going down.

All of the art in the game was created by Dragomir Ivanov.

Sknat

Platform(s): Any

Technology: Java

Duration: 11 weeks

Description:

Sknat is a multiplayer 2D Shooter game. It can be played in 1v1, 2v2 or 1vAI mode. The goal of the game is to collect a key, which is randomly spawned in the map and bring it to the missile launcher panel in the middle of the map. When the missile launcher is activated by a team, it launches a missile to the opponents base. The goal is to destroy the opponents base. Each player controls a tank and can shoot at other players.

The game was developed as part of the Team Project module during my 2nd year at the University of Birmingham. I worked in a team of 5 people. A colleague of mine and I were working together and our responsibility was to implement the networking and the AI of the game. The others were responsible for the UI and the gameplay.

The game and all components were written in Java as required by the university.

The networking part was implemented using UDP and TCP. The game produced various messages, containing information about the updates in the game state. The messages usually contained multiple feature update, separated by a special character to reduce the size and the number of sent messages. The game followed a server-client architecture with one of the clients creating a server, connecting to it and allowing other players to join as well. Important information such as the game settings, collection of a key, missile launch, a player shooting, etc, was transferred to the server using TCP and then transmitted to all other clients using TCP as well. Not so important information which was generated frequently and did not have a big impact if lost, such as the player's position and the rotation of the gun, was transmitted through UDP to the server and through UDP to the clients.

The AI was kept relatively simple. It has a few states based on the game state and every state is a goal where the tank should go. Finding a path through the map was done by breaking up the map into small sectors and using the A* search algorighm to find the shortest path to the goal. The AI was also able to shoot in the players direction based on some conditions.

Onet Connect Animal - Cava

Platform(s): Android

Technology: Unity 5

Duration: 3 weeks

Description:

Cava is an Onet Connect type of game. The goal is to find matching pairs of tiles which can be connected with 3 or few lines and clean the board in the given time limit. Once all tiles have been removed, the player progresses to the next level, where a new random field is generated and the time limit is reduced. If the time limit runs out or there are no possible moves the player has lost the game and has to start again from the beginning. The game also has an offline leaderboard.

I worked on the game as a freelancer and the game was created by the idea and requirements of the client.

The game was created using Unity 5. I was responsible for all parts of the development of the game as I was the only one working on the game. The backend generates the puzzles, checks if a move is valid, finds the next possible move to return as a hint, etc. I also had to create the user interface as per design documents, which involved the creation of some graphics in GIMP. The client also wanted the Unity project to be easily customisable, so I made the game as modular as possible. The project allowed to modify and customize almost all parts of the game (icons, size of the field, music, time limit, progression rate, etc) through the editor, requiring no coding skills.

All scripts were written by myself in C# apart from the AdSense module. Most of the graphics in the game, excluding the tile background and some of the menu icons, were provided by the client.

The project's duration was three weeks, during which I worked a few hours a day. At the end of each day I was producing a prototype, which was then reviewed by the client and feedback was recieved.

Distributed & Parallel Computing Project

Platform(s): Any

Technology: CUDA, OpenGL

Duration: 3 weeks

Description:

A project that was part of the Distributed And Parallel Computing Module. It's goal is to generate and rotate a torus.

The torus is generated by sweeping a circle around another circle and recording the points during the sweeping. The generated torus has 500,000 points, which are generated in 140ms.

The rendering is done through OpenGL. All lighting calculations are done by recalculating normals every rotation, as well as rotation of the position of the points. The torus is rendered using Phong lighting with a light source directly above the torus. The camera remains in the same position and looking towards the origin where the torus is located.

The CPU implementation averages 10 fps, where the CUDA optimised implemetation averages 21 fps. In comparison, rendering the same object using OpenGL without making any computations achieves 30 fps.

Offroad Appocalypse

Platform(s): Windows Phone

Technology: Unity 5

Duration: 2 weeks

Description:

An infinite runner game. The armored vehicle moves along the road and has to avoid rocks on the road by changing to one of the three lanes. Switching lanes is done by swiping left or right. If a rock is hit, the player loses one of his three lives. There are car parts on the road, which give extra lives. The difficulty increases with time and the goal of the game is to reach as far as possible.

The game was created in Unity 5 as part of the second round of the Rising Star competition. I was given a very simple and basic template of a game and my task was to make the game fun, add features and realease the game to the Windows Phone market in two weeks.

All scripts were written in C# and I was responsible for the gameplay, as well as the user interface of the game. Most 3D models, excluding the rocks, are free assets from the Unity Marketplace.

Square Remover

Platform(s): Android

Technology: HaxeFlixel

Description:

A very challenging puzzle game. The goal of the game is to clear the board. Starting from the initial middle empty space the player has to choose one color and all pieces of that color that are reachable from the initial position will be removed. For each level, the player has a limited number of moves in which the field have to be cleared. If the player fails, he has to retry the level. The game consists of 500 levels increasing in difficulty.

The game was created using the HaxeFlixel framework. I worked together with Teodor Mihov on the game. He was responsible for creating the User Interface. I created the backend in C++, which contained the game logic and was able to generate puzzles, generate their solutions, evaluate their difficulty, order them according to it and produce an output file that contains all the levels' information. The game logic was then translated to Haxe by Teodor and was implemented in the actual game.

All art assets were created by Teodor.

Project RH

Platform(s): Windows

Technology: Unreal Engine 4

Description:

Project RH was a realistic horror game project. We were a team of 1 artist, 1 producer and me as the single programmer in the team.

I was responsible for the gameplay and interaction with the environment. Everything was written in C++ with the exception of some very simple features that were implemented using Blueprints. As the artist and the producer had to be able to tweak the features, I tried to make everything as modular as possible and allow most of the settings to be configured in the editor, without the need for writing any code.

The project was put on halt due to us not being able to spend any time on the project no more.

Energon

Platform(s): Windows Universal Platform

Technology: Unity 5

Duration: 2 weeks

Description:

An infinite runner game. The goal of the game is to collect Energon spheres that are the same color as the color of the hull of the player's ship. If a sphere with the same color is missed, the game is over. The same happens if a sphere of different color hits the ship.

The game was created in Unity 5 as part of the second round of the Search For A Star competition. I was given a very simple and basic template of a game and my task was to make the game fun, add features and realease the game as a Universal Windows Platform App to the Windows Store in two weeks.

All scripts were written in C# and I was responsible for the gameplay, as well as the user interface of the game. The 3D ship model as well as the icons in the user interface are free assets. The Energon spheres are made of a simple sphere combined with a custom GLSL shader in order to create the "electric" effect and modify the colors of the spheres during runtime. Another GLSL shader was created for the player's ship, which displays the current and next target colors.

The game also has an Online Leaderboard, which was implemented as a REST Server written in Node.js.

Another feature of the game is In-Game Purchases using the Windows Store. This invlolved creating a separate Dynamically Linked Library written by me in C# that performs the connection between the Unity game and the Windows Store.

Opponent Modelling in StarCraft Brood War

Platform(s): Windows

Technology: BWAPI, C/C++, Python

Duration: 11 weeks

Description:

The project aims at implementing two different approaches for opponent modelling in StarCraft Brood War, as well as an evaluation system in order to show the algorithms’ performance. The first approach is a Bayesian Network, which tries to predict a binary value for all types of buildings and units indicating whether or not the opponent has created any of those. The second approach makes the opponent model more detailed by implementing a Neural Network, which tries to predict the exact number of buildings and units the player’s opponent has. A system for gathering and extracting the required data in the appropriate format, evaluating the prediction algorithms’ performance and automating the parameter learning of those algorithms has also been created as part of this project.