Skip to main content

Faculty of Science Project Summaries

Supervisors  

Cristiano Politowski| Nisha Agarwal |

 

Supervisor name: Cristiano Politowski

Project title: Autonomous Playtesting: Deep Reinforcement Learning Agents for Automated Game Quality Assurance

Summary of research project: Games are complex software, yet gameplay quality assurance (QA), finding bugs, softlocks and balance flaws, is largely manual and costly. This project builds autonomous playtesting agents driven by deep reinforcement learning (DRL): software that learns to play a game and report defects. The student will (1) build engine-side adapters exposing actions and game state for open-source Godot and Unity games, (2) train DRL agents under player personas (speedrunner, explorer, completionist), and (3) measure whether the resulting playtraces expose known bugs, unreachable areas and difficulty spikes. Deliverables: an open-source toolkit and a reproducible benchmark of agent-detected defects.

Student responsibilities/tasks:

  • Implement engine-side adapters (Godot, Unity) exposing game actions and state to an external agent.
  • Train and tune deep reinforcement learning (DRL) agents for 2-3 player personas.
  • Run experiments on open-source games and log playtraces.
  • Analyse traces to flag bugs, unreachable areas and difficulty spikes.
  • Document results in a public repository.
  • Weekly meetings, code reviews and a final report are included.

Student qualifications required:

  • Completed second year of a Computer Science or Software Engineering program; minimum B average.
  • Proficient in Python or C++; familiarity with game development and game engines.
  • Coursework in machine learning or artificial intelligence (AI) an asset.
  • Experience with version control (Git) and Linux.
  • Interest in game development and testing.

Expected training/skills to be received by the Student:

  • Training deep reinforcement learning (DRL) agents: reward design, policy training and hyperparameter tuning on real game environments.
  • Game engine programming: building Godot and Unity plugins that expose game state and controls through a clean programming interface.
  • Empirical research methods: formulating hypotheses, designing experiments, logging data and analysing results with statistical rigour.
  • Software engineering practice: version control, code review, reproducible experiment pipelines and open-source release of the toolkit.
  • Scientific communication: writing a technical report, presenting to the research group and co-authoring a potential conference paper.

Length of award: 14 Weeks

Location of award: Hybrid

Available Award: NSERC USRA

 

Supervisor name: Nisha Agarwal

Project title: Quantum Treatment of Tip Enhanced Raman Scattering Modes

Summary of research project: Tip-Enhanced Raman Scattering combines scanning probe microscopy with surface-enhanced Raman scattering to analyze chemical processes at the nanoscale. A laser focused on a sharp, metallic probe tip generates a strong localized electromagnetic field (a "hotspot"). When this hotspot is brought close to a sample, it significantly enhances the Raman signal, allowing for chemical imaging with resolution down to angstroms. Our goal is to analyze the field of a laser beam around a metal-coated tip and study the associated skin effect and surface plasmons. This dynamics of a simple molecule in this field will be studied as an open quantum system to quantify the effect of thermal and quantum noise.

Student responsibilities/tasks:

  • The intern analyzes molecular structures and vibrational symmetries, builds a simplified Tip‑Enhanced Raman Scattering (TERS) lossy‑cavity model, and simulates polarization‑ and frequency‑dependent Raman contrast.

Student qualifications required:

  • Completed courses in Quantum Mechanics (QM) and Electromagnetism (EM).
  • Experience with Density Functional Theory (DFT) or molecular modeling.
  • Basic coding skills (Python or MATLAB).
  • Strong analytical and visualization skills.
  • Minimum A- average in physics courses.

Expected training/skills to be received by the Student:

  • Training in quantum‑chemistry modeling (normal modes, symmetry, Raman tensors) and visualization of molecular vibrations.
  • Development of a lossy‑cavity theoretical model to understand near‑field enhancement and polarization‑dependent Raman contrast.
  • Skills in data analysis, comparing theoretical predictions with experimental TERS measurements to interpret vibrational behavior at the nanoscale.

Length of award: 14 Weeks

Location of award: In-Person

Available Award: NSERC USRA