Philippe Teichmann Projects Portfolio

​Commercial game. A roman survival city-builder/colony-sim. Demo out, 92% positive with 200+ reviews. [link]

​

​Developed entirely by me. Some noteable elements:

• Implemented custom A* pathfinding (multi-threaded highly performant solution)

• Developed a customizable procedural environment generation algorithm, systems with individually highly-performant simulated citizens, enemy AI and controllable armies and much more (serialization system, economy, research system, optimized humanoid animations etc.).

• Independently developed a commercial product with tens of thousands of users (demo released, full game coming soon), handling all gameplay systems, intuitive UI/UX, AI, optimization, and iterative improvements. Written over 100,000 lines of code.​

• Developed an MLP to classify the MNIST handwritten digits dataset, achieving competitive accuracy.

• Designed and evaluated a TensorFlow CNN with varying layer configurations and hyperparameters, comparing performance against the MLP.

• Leveraged scikit-learn and pandas for efficient data pre-processing.​​

Code: https://colab.research.google.com/drive/1GpBi0tNFXLun23fmwVl0OOOuhx_RAiRI?usp=sharing

​

Results, Analysis and Discussion: https://drive.google.com/file/d/1Ztk0BQLzslB4E88bka__gWVv93E1I4kM/view?usp=sharing

A turn-based strategy digital board game. Developed in Unity entirely by me (art, game design, programming), apart from the use of some VFX assets and SFX assets. 

​​

• Local and Online Multiplayer (P2P)

• Singleplayer against AI (minimax, pruning)

• Level editor/creator (play and share levels you make), competitive mode, tutorial mode, steam achievements etc..

• Iterative development with a testing QA team (same as with Roman Triumph)

​

https://store.steampowered.com/app/1517240/CHESS_with_LASERS

​

​

Started as a baseline functionality prototype for a stealth FPS hitman game​.

​

FPS controller with realistic weapon animations done by a contracted artist, sounds and visual effects.​

Weapon customization system with realistic attachments​ which modify behavior of the weapon.

Focus on creating good game feel​ and realistic gunplay through detailed player feedback (shell casings, smoke etc.)

​

Enemy active ragdoll for realistic hit response with realistic blood visual effects.

​

​

• Created a high-performance multi-pathogen agent-based model for simulating the spread of multiple pathogens (e.g., SARS-CoV-2, Influenza, HRS, RSV) within a synthetic population, capturing pathogen-pathogen interactions in hosts. 

• Developed in Python with use of SciPy, Numba and NumPy libraries and data-oriented design for high performance computing. Developed a theoretical model and implemented it.​

​Github: https://github.com/PhilippeTei/pathosim

Research project at McGill University under Professor Caroline Wagner (Bioengineering, Summer 2023–Fall 2023). An academic paper is in progress.

Architecture of the model (pathosim) [link]

Poster which provides an overview of the research:

​

10 weeks game development competition in an 8-person team.
- Built multiplayer networking for player interactions.
- Developed the AI, animation, UI and gameplay mechanics.

​

Our game was nominated by Ubisoft for one of the awards. 

​

Video: https://www.youtube.com/watch?v=ue97fEyqtLw

​

Note: the team has proceeded to building the game fully without me. The linked video was filmed using the competition's build.

A simulation and visually representation of the spread of SARS-COV-2 in a store, according to user-specified simulation parameters. A general mathematical framework for estimating risk of airborne transmission of SARS-COV-2 given by the American Institute of Physics was implemented in a 3D real-time virtual setting populated with artificially intelligent shoppers.

Developed in Unity entirely by me.

Makes use techniques such as pooling, cpu time slicing, A* pathfinding, hash grids, finite state machines and more.

Explanation of  model [link]

• An automated water pumping system for feeding water in indoor plants and trees. Comprised of an Arduino board, a capacitive water level sensor, a water pump and a LED display.  

• A meccanum wheel robot equipped with an ultrasonic sensor capable of avoiding obstacles and moving in all direction. This was done with Lego Mindstorm.