Computer science (CS) departments rely heavily on graduate teaching assistants (GTAs), yet many departments struggle to provide effective pedagogical training, particularly for international GTAs who face additional cultural and communication challenges. While pedagogical training improves student outcomes, engaging GTAs with diverse career priorities in such training remains difficult, and resource constraints often lead departments to default to less engaging online formats. This experience report examines the implementation of a low-cost, in-person training program for CS GTAs using a flipped classroom approach that supplemented existing online modules. Our cohort of 34 international GTAs actively engaged with the training, which focused on grading practices, feedback techniques, cultural competencies, and common teaching scenarios. Survey data collected before and after training and midway through the semester revealed that participants found the training highly beneficial, with 94% reporting increased preparedness for their teaching roles. GTAs consistently applied learned skills throughout the semester, particularly in providing effective feedback (90%) and using rubrics (76%). The program's structure-requiring minimal faculty resources while yielding significant improvements in GTA confidence and teaching practices-offers a practical, replicable model for other CS departments seeking to enhance GTA preparation without substantial resource investment.

Computer science (CS) relies heavily on teaching assistants (TAs) who are often untrained in CS pedagogy. Existing research on CS TA training typically studies American undergraduate TAs at high-resource universities, ignoring the many universities that use graduate TAs, who are often international students, and that don't have the resources to implement the training strategies discussed in the literature. We describe our approach to implement graduate TA training in a high-diversity, low-resource context. We present a needs assessment, design, pilot test, and deployment of our training course, and discuss implications for other similar departments hoping to train their TAs.

Teaching assistants (TAs) for introductory computer science courses are most often responsible for providing feedback on student code. TAs, however, lack teaching experience and are rarely trained in how to give effective feedback that positively impacts student learning. The lack of training is particularly problematic when TAs are asked to give feedback in real time, e.g. during in-class coding exercises. We analyzed data from multiple semesters, where CS1 TAs and instructors provided written feedback on in-class coding exercises. Importantly, a very small percentage of feedback met our gold standard for high quality. This finding reveals a need for training TAs to provide more effective feedback in introductory programming courses.

One way to introduce active learning in large introductory computer science courses is for students to solve coding exercises in class. Although it is commonly understood that re-solving a problem after receiving feedback can deepen understanding and improve performance, students often do not have opportunities to make multiple attempts on in-class exercises due to practical classroom constraints in time and logistics. In this experience report, we share the results from our experience with multiple attempts in our CS1 course of 114 undergraduate students. In each of 2 lectures on arrays, students were given two in-class coding problems. The first was a practice problem, where they had either one attempt or two attempts to solve the problem, and the second was a test problem where all students had only one attempt. We measured how having one attempt or two attempts on the practice problem impacted student performance on the test problem. We observed that students who used a second attempt to try re-solving missed practice problems were more likely to succeed on the test problem, even if they missed both tries on the practice problem. This work suggests that, given the right context and tool, multiple attempts on in-class exercises in CS1 might improve student performance.

In large programming courses, it can be difficult for instructors to identify students who need help. Often the earliest indication of trouble is when a student fails an exam, which unfortunately can be too late. Using data from 7 sections of CS1 over multiple semesters, we show that performance on lab and in-class coding exercises can be used to accurately predict which students will fail or struggle on upcoming weekly lab assignments. We found that recent relevant in-class coding exercises were the best features for building accurate models. This approach has potential in helping CS1 instructors identify students who need help, determine which topics need additional attention, and formulate intervention plans, all on a weekly basis before each lab meeting.

Instructors often implement active learning in CS1 by giving students in-class coding problems. Students need feedback on their work to improve. While some systems provide automated feedback, human feedback is more effective for novice learners. However, instructors cannot provide feedback quickly at a large scale. Peer feedback systems help students get prompt feedback during class. Existing CS peer feedback systems usually support feedback on completed code rather than work in progress, which limits opportunities to reflect on the feedback and correct their work. We introduce a novel system for giving peer feedback on code in progress during CS1 classes, as well as a pilot test of the peer feedback process in CS1. Our initial experience has implications for the delivery of in-class instruction and for teaching growth mindset in order to take full advantage of peer feedback.