George Daneliuc
Building the future through data engineering, analysis, software development and research.
I am passionate about how we can use technology and AI to solve modern problems.
Projects
Gridless Pathfinder in OpenRA
The first working version of gridless pathfinding for OpenRA, an open-source port of the classic RTS game Red Alert. I developed this as part of a game project I was undertaking called Jire. This uses a modified Any Angle A* algorithm called Theta*, porting code developed by Tansel Uras in C++ at http://idm-lab.org. Paper on the topic is T. Uras and S. Koenig, 2015. An Empirical Comparison of Any-Angle Path-Planning Algorithms. In: Proceedings of the 8th Annual Symposium on Combinatorial Search
Jire
The continuation of the game project that began in OpenRA, now being built in a custom engine built within the Raylib library for increased performance. Uses a variation on the Theta* pathfinder from the "Gridless Pathfinder in OpenRA", incorporating a Constrained Delaunay Triangulation algorithm from Alejandro Villalba that I have manually ported to C++. Also uses a hybrid data structure that combines a MinHeap with a Hash table to achieve a pathfinding time of 0.5 ms, which is approx. 100x faster than the C# version built within OpenRA. A video of a 3D version I have now made of this is now available via this link. This version is closer to the final product, as the game will take place in 3D.
PONG using Observables
Built a game of Pong in JavaScript using Reactive programming. Reactive programming is an advanced programming paradigm seen in the JavaScript library RxJs, used for improving concurrency, thread management and asynchronous programming. My game includes many additional features, such as 200 levels that progressively increase in difficulty, multiple balls being in play at higher difficulties, and an AI that can handle all this and is almost impossible to beat!
Honours Thesis, Monash University
Used the latest transport simulation software (MATSim and SUMO) to simulate car use changes within Melbourne, Australia after introducing autonomous vehicles. These autonomous vehicles had two characteristics: added passenger comfort that slows the car down by limiting acceleration and deceleration, and the ability to perform other activities while driving. While these are both positive effects, it was hypothesised the limits on acceleration and deceleration would add congestion, reducing the appeal of car use. The simulation software was used to test this theory, to see whether car use changed as a result and how.
Education
Bachelor of Software Engineering (Honours)
Degree Grade: Distinction Average
Honours Research Grade: High Distinction Average
Certificates
Machine Learning by Andrew Ng
Online Course
Grade: 94%
Skills
Data
Languages
- Python
- R
- SQL
- Excel
Software
- QlikView
- Power BI
- Jupyter Notebooks
- Microsoft SQL Server
- Microsoft Access
- Microsoft SSIS
- Microsoft Excel
- R Studio & R Core
- Azure Machine Learning
Software
And Web
Architectures
- React
- Bootstrap
- Git
- Azure DevOps
- RESTful APIs
- NodeJS
Backend Languages
- Python
- C#
- Java
- C++
- JavaScript
- XML
Frontend Languages
- HTML5
- CSS3
- JavaScript
Other Capabilities
- Graphic Design
- Photo Editing
Research
Transport Simulators
- MATSim
- SUMO
Machine Learning Libraries
- Word2Vec and GloVe
- Natural Language Processing Toolkit (NLTK)
- Weka
Management
Strengths
- Project Management
- Process Development
- Coaching and support
- Documentation
Interests
TBC