For students learning to write code, developing strong foundational coding skills and cultivating proper code style early on is crucial for success in subsequent courses and professional work. TA-Bot, an automated assessment tool, incorporates novice-friendly style suggestions wrapped around an industry-standard static analysis tool, code correctness testing, and an innovative rate-limiting system called Time Between Submissions ("TBS''). This system works in conjunction with a gamified incentive mechanism designed to motivate students to start weekly assignments earlier. Our hypothesis posited that this incentive, when combined with the inherent effects of TBS, would not only encourage students to initiate assignments sooner but would also prompt them to address more style-related issues and produce higher quality code.

This innovative practice full paper describes the effectiveness of self-hosted large language models (LLMs) in assisting with the automatic grading of CSI assignments. Educators often rely on automated review of student code submissions in larger courses. Despite recent advancements, current systems primarily focus on assessing functionality, with important aspects such as code structure, efficiency, and style often relegated to secondary foci. LLMs provide an increasingly attractive addition to these systems to enhance those overlooked areas. Prior research has shown LLM's capable of assisting students in understanding and resolving programmer error messages, correcting syntax errors, providing enhanced explanations of code segments, or even generating code. The absence of freely available, purpose-designed LLMs for grading and providing feedback on code submissions prevents widespread adoption by educators. Remotely-hosted systems, such as fine-tuned GPT models, have shown promise, yet the associated risks of privacy breaches, ethical considerations, and recurring costs make this approach unfeasible as a universal solution. To mitigate these concerns, self-hosted open-source models are an alternative that can operate on consumer-grade hardware and prevent some privacy and security concerns. While no purpose-built solution yet exists, it is unclear if any existing models are powerful enough to facilitate automated grading. To explore these questions, we present a two-phase analysis, leveraging real grading data from a semester length, introductory CSI course with 124 students and nine programming projects. Nine stable LLM models were selected and repeatedly prompted to grade student submissions using the same context that a human teaching assistant (TA) was given. This paper analyzes 1,172,383 API requests, totaling 33.4 days of active runtime, evaluating model consistency, ability to adhere to specified constraints, and comparison to human-generated grades. The results show various models' inability to consistently grade assignments, albeit with some exceptions. The importance of providing comprehensive context to models was highlighted, as incomplete contexts resulted in worse performance. Other models struggled with longer prompts, delivering less consistent results. Despite disparities between AI-generated and human-assigned grades, the potential for refinement is clear; improved rubrics or selective fine-tuning could enhance model output. Future work will focus on analyzing models' qualitative justifications for grades, refining rubrics, training on domain-specific datasets, and fine-tuning the highest performing models to potentially improve grading accuracy.

RISC-V is an open computer architecture that has gained increasing popularity in recent years. Companies such as Google, Nvidia, and Huawei have all announced or developed CPUs based on the RISC-V architecture. The increasing popularity of RISC-V along with its simplicity make it an ideal platform for students to learn low-level operating system concepts. We have ported Embedded Xinu, a simple, lightweight, and education-focused operating system, to a baremetal RISC-V board. Embedded Xinu has been used to teach thousands of students operating systems over the past two decades. This new port is the first education-focused operating system designed to run on baremetal RISC-V. In the following sections, we describe the challenges in porting Embedded Xinu to support the RISC-V architecture. We describe how practitioners can adopt Embedded Xinu to teach low-level CS systems courses such as operating systems. Finally, we reflect on our experience using Embedded Xinu on RISC-V to teach operating systems in Spring 2023.

Automated Assessment Tools (AATs) have been used in undergraduate CS education for decades at many universities. TA-Bot, a modular AAT, has existed in some form for 25 years serving thousands of students across multiple universities. Previous research has shown that the earlier students start assignments, the better scores they receive. TA-Bot implements a novel dynamic rate limiting system to incentivize earlier student submissions. As the assignment deadline approaches, the cooldown from when a student submits to when they can make their next submission increases. Thus, students who start earlier are given more opportunities for automated feedback than a student who started closer to the deadline. The experiment discussed used TA-Bot over two semesters involving 144 students in CS1. When the dynamic rate limiting was enabled, students tended to start assignments earlier.