Chee Wei Tan

Project-based teamwork is central to computing education, yet opportunities for students to practice leadership, particularly in realistic and low-risk environments, remain limited. To address this gap, we propose a simulation framework that leverages large language models (LLMs) to generate dynamic, role-based team interactions for situated leadership training. Each LLM assumes a distinct software role (e.g., frontend, backend, quality assurance), while the student acts as team leader, coordinating the group through natural-language dialogue. Rather than relying on pre-scripted scenarios, the simulation elicits organic negotiation and decision-making as agents surface competing priorities and constraints. Even within a short planning iteration, students are required to resolve ambiguity, mediate conflicts, and align the team toward shared goals. This work demonstrates that multi-agent LLM simulations can serve as a powerful tool for experiential leadership development in computing contexts and provides a scalable foundation for future extensions involving iterative implementation and evolving project artifacts.

Learning competitive programming and advanced data structures presents a significant challenge for students due to the high cognitive load and the need for long-term concept retention. Traditional online judge platforms provide a static, one-size-fits-all experience that fails to adapt to individual student skill levels or address the exploration-exploitation tradeoff inherent in learning. In this paper, we introduce MyCodeWeapon, a novel learning platform that provides a personalized and adaptive educational environment for competitive programming. MyCodeWeapon integrates three algorithmic components: (1) a dynamic difficulty calibration model using an Elo rating system with an adaptive K-factor to jointly estimate student ability and problem difficulty; (2) a contextual multi-armed bandit based on Thompson Sampling for adaptive problem recommendation that balances exploration of new topics with exploitation of existing skills; and (3) a spaced repetition scheduler inspired by the FSRS algorithm to promote long-term concept retention. We describe the architecture of the MyCodeWeapon platform---built on Next.js, Supabase, and Judge0---and report a pilot study showing that the adaptive group achieved higher post-test scores, attempted more problems, and demonstrated greater learning efficiency. This work aims to demonstrate a more effective and engaging paradigm for scaling programming education.

This paper introduces a deep reinforcement learning-based block coordinate descent (DRL-based BCD) algorithm to address the nonconvex weighted sum-rate maximization (WSRM) problem with a total power constraint. Firstly, we present an efficient block coordinate descent (BCD) method to solve the problem. We then integrate deep reinforcement learning (DRL) techniques into the BCD method and propose the DRL-based BCD algorithm. This approach combines the data-driven learning capability of machine learning techniques with the navigational and decision-making characteristics of the optimization-theoretic-based BCD method. This combination significantly improves the algorithm's performance by reducing its sensitivity to initial points and mitigating the risk of entrapment in local optima. The primary advantages of the proposed DRL-based BCD algorithm lie in its ability to adhere to the constraints of the WSRM problem and significantly enhance accuracy, potentially achieving the exact optimal solution. Moreover, unlike many pure machine-learning approaches, the DRL-based BCD algorithm capitalizes on the underlying theoretical analysis of the WSRM problem's structure. This enables it to be easily trained and computationally efficient while maintaining a level of interpretability. Through numerical experiments, the DRL-based BCD algorithm demonstrates substantial advantages in effectiveness, efficiency, robustness, and interpretability for maximizing sum rates, which also provides valuable potential for designing resource-constrained AI-native wireless optimization strategies in next-generation wireless networks.

Learning to learn is becoming a science, driven by the convergence of knowledge tracing, signal processing, and generative artificial intelligence (GenAI) to model student learning states and optimize education. We propose CoTutor, an AI-driven model that enhances Bayesian knowledge tracing (BKT) with signal processing techniques to improve student progress modeling and deliver adaptive feedback and strategies. Deployed as an AI copilot, CoTutor combines GenAI with adaptive learning technology. In university trials, it has demonstrated measurable improvements in learning outcomes while outperforming conventional educational tools. Our results highlight its potential for AI-driven personalization, scalability, and future opportunities for advancing privacy and ethical considerations in educational technology. Inspired by Richard Hamming’s vision of computer-aided “learning to learn,” CoTutor applies convex optimization and signal processing to automate and scale up learning analytics, while reserving pedagogical judgment for humans, ensuring that AI facilitates the process of knowledge tracing while enabling learners to uncover new insights.