Welcome to the public overview of my OMSCS graduate work at the Georgia Institute of Technology.
My focus has been on quantitative modelling, machine learning, and AI, combined with a multidisciplinary lens that includes cognitive science, ethics, finance, and digital strategy.
This repository serves as a high-level index of my 10 OMSCS courses and their associated projects. For a more personal, narrative view of the journey, including how the first seven courses fitted together and what it was like balancing OMSCS with National Service reservist training, see my blog post. The concluding finale post will follow once I have survived my overseas ICT by June. After all, whether this regional graduation trip counts as a curse or a blessing is still an open question.
All project code resides in private repositories to comply with Georgia Tech’s Academic Honor Code, but is available for reference upon request. Each of the courses below corresponds to one or more private repositories containing:
- Full source code and project structure
- Notebooks, experiments and analysis where applicable
- Reports and documentation describing goals, methods and findings
- Quantitative modelling and systematic strategy design
- Machine learning, AI, and NLP
- Human cognition, ethics, and decision-making in high-stakes domains
- Software engineering practices for reliable, maintainable systems
Applied machine learning and reinforcement learning to systematic trading strategies and portfolio management.
- Algorithmic Trading System Capstone: Built an automated trading engine that trained a Random Forest ensemble on custom technical indicators (e.g. MACD, RSI, Bollinger Bands, Stochastic Oscillator) to forecast daily long/short signals and drive end-to-end order generation and backtesting.
- Developed and evaluated rule-based and ML-driven strategies using an event‑driven backtesting framework, measuring performance via Sharpe ratio, drawdown and benchmark-relative returns.
- Experimented with a portfolio optimiser under varying risk controls and position sizing rules to balance risk and return.
Built end-to-end financial models to support valuation and capital allocation decisions.
- Designed spreadsheet-based and programmatic DCF, scenario and sensitivity models to assess project and firm value under uncertainty.
- Incorporated capital structure, cost of capital and risk assumptions into model design to evaluate investment decisions.
- Structured models to be auditable and parameter-driven, enabling clear communication of assumptions and outcomes to non-technical stakeholders.
Implemented core AI algorithms from first principles to solve search and decision-making problems.
- Implemented A*, minimax and alpha–beta pruning to solve deterministic and adversarial search tasks efficiently.
- Built probabilistic reasoning systems using Bayesian networks and decision-theoretic agents for inference and action under uncertainty.
- Analysed algorithmic complexity and performance trade-offs when scaling to larger state and action spaces.
Explored decision-making and control under uncertainty in simulated robotic environments.
- Implemented localisation and path‑planning algorithms that combined probabilistic state estimation with cost‑aware trajectory planning.
- Designed controllers that balanced path optimality, stability and robustness to sensor noise and actuation error.
- Gained practical experience with sequential decision problems where state, action and dynamics interact over time.
Investigated how knowledge representation shapes intelligent behaviour and explainability.
- ARC‑AGI Agent Capstone: Engineered a modular, transformation‑based agent for the ARC‑AGI visual reasoning benchmark, using a domain‑specific language and libraries of spatial and colour transformations to solve abstract grid‑based puzzles from examples.
- Designed knowledge‑based agents using rule‑based systems, case‑based reasoning and explanation‑based learning across varied problem domains.
- Analyzed alternative representations (frames, semantic networks, rules) to assess how they affect problem‑solving efficiency and transparency.
Modelled and analysed human language for classification, prediction and sequence tasks.
- Question Answering Agent Capstone: Built a retrieval‑based question answering system that indexed a document collection and answered free‑form natural language questions by ranking candidate passages and extracting concise responses.
- Implemented text classification, language modelling and sequence tagging models using both classical and neural approaches.
- Applied vector representations and sequence architectures to capture semantic and syntactic structure, and evaluated models with task‑appropriate metrics whilst inspecting error patterns to improve robustness.
Examined the broader impact of AI systems and how to design them responsibly.
- Mortgage Lending Fairness Capstone: Built an end‑to‑end pipeline on 2023 HMDA mortgage data for major US banks in Fulton County, GA, exploring bias in approval decisions by protected characteristics such as race and sex.
- Applied re‑weighting after exploratory data analysis to mitigate dataset bias, and compared fairness–performance trade-offs across multiple models.
- Analyzed case studies of algorithmic harm and developed principles for deploying models that balance accuracy with accountability, governance and regulatory expectations.
Learnt how humans perceive, reason and decide across multiple disciplines.
- Studied theories of memory, attention, learning and decision‑making from psychology, neuroscience and AI.
- Connected cognitive models to interface and system design, focusing on how people understand and interact with complex information.
- Applied these ideas to think about human‑in‑the‑loop systems and how model outputs are interpreted by decision‑makers with varying levels of expertise.
Applied data‑driven methods to customer behaviour and digital channels.
- Examined customer segmentation, funnel analysis and experiment design to evaluate marketing strategies and channel effectiveness.
- Leveraged data to inform campaign decisions, measure uplift and quantify the impact of different messages and touchpoints.
- Practised translating quantitative findings into clear, actionable recommendations for commercial stakeholders.
Developed software engineering discipline around building and maintaining complex systems.
- Individual projects: applied object‑oriented design, refactoring and automated testing to keep growing codebases maintainable over time.
- Group projects: collaborated in an agile team to design and implement a full‑stack application with version control, code review and CI/CD.
- Emphasised communication, documentation and code clarity so that systems remained understandable and extensible for other engineers.
- To request a walkthrough of particular projects or discuss opportunities in specific roles, please reach out via the contact information listed on my resume.