Teaching Statement
Teaching Philosophy
My approach to academic teaching is fundamentally rooted in my background in competitive programming, first as a participant and later as a coach. In 2007, I started my coding journey and I was fortunate to win a medal at the national level in 2008, followed by a silver medal in IOI 2010, one of the most prestigious coding competitions internationally. In 2008, I led the coding club in my high school, often coaching my friends and juniors. This continued during my undergraduate studies, where I trained peers and juniors specifically for competitions.
Training for a competition is intensive and often requires long periods of learning. This experience taught me that engaging teaching is a must; maintaining students’ motivation is key to better learning outcomes. In competitions, our goal is for the student not only to know high-level ideas but to deeply understand the algorithm so that they can actually implement it.
This paradigm carries over to my academic teaching. I try to deep-dive into topics rather than just giving a high-level overview. For example, I deep-dive into the formula and the hands-on implementation of the subjects that I teach. I want students to be able to employ the algorithms, not just discuss them.
Finally, I ensure that the teaching is interactive and personalized to engage more with the students. I often ask questions to benchmark understanding and adapt if topics have not yet been mastered. I check the student’s status and every so often provide extra sessions for those who need more help, such as extra tutorials in coding or PyTorch.
MBZUAI Teaching
NLP801 - 2023 I joined MBZUAI right when the NLP department was formed and was appointed as the main instructor for NLP801, an introductory course for PhD students. Since the course was new, I had to design it from scratch, with the help of Thamar Solorio. As the program was new, we only had 6 students. I designed the course with the assumption that the PhD students were already familiar with NLP basics; hence, it was more advanced from the get-go and focused on discussions of current trends. However, along the way, I noted that some students had zero background in NLP, which required me to adjust the teaching on the go. This provided a learning opportunity to improve my upcoming teaching.
NLP702 - 2024 In the next term, I was tasked to co-instruct NLP702, a master-level course on advanced NLP covering modern topics such as transformers. However, I was not the lead instructor and taught only a minority of the sessions. Note that at MBZUAI, courses are often taught by multiple lecturers. Unfortunately, this course received mixed feedback. While students expressed concerns about the depth and repetitiveness of the material in the broader course, the feedback for my specific sessions was excellent. In fact, the course evaluations explicitly suggested that I should take on a larger portion of the teaching load.
NLP702 / NLP806 - 2025 In the next term, I was trusted to take the lead on NLP702. The class was combined with PhD students (NLP806), so the class size was even larger. I updated the materials to remove redundancy and to include more recent advancements, such as State-Space Models, and more practical advancements in distributed training of Large Language Models, interpretability, and more. I also taught the majority of the sessions. Feedback improved significantly from the previous year and the students were generally positive about the course. My personal teaching feedback also improved.
Teaching Feedback
The following is my course feedback, and it shows that the students gradually increases. It also shows the improvement from NLP702 after I took the lead and improve the course. I also highlight positive feedback, including comments that students submitted in the negative feedback section.
| Course | Term | Size | Course Rating (out of 10) |
Lecturer Rating (out of 5) |
University Avg |
|---|---|---|---|---|---|
| NLP801 (main lecturer) | Fall 2023 | 6 | 8.00 | ? | |
| NLP702 (secondary lecturer) | Spring 2024 | 26 | 7.20 | 4.70 | ? |
| NLP702 (main lecturer) | Spring 2025 | 35 | 8.46 | 4.55 | 4.39 |
| NLP806 (main lecturer) | 16 | 4.75 | 4.39 |
| Course | Feedback |
|---|---|
| NLP801 (2023) | The instructors helped accelerate the learning curve for the taught students, insuring that all students catch up to the material that is being taught |
| NLP702 (2024) | Dr. Alham’s part especially during his session’s labs because we learn everything from scratch and this is very helpful to enhance the knowledge practically and theoretically. |
| NLP702 (2024) | Lectures of Dr. Alham! They were very useful, very clear, very interesting and relevant. For example, he was the first lecturer in my life who explained how GPU’s work when we train the model, etc. |
| NLP702 (2024) | Outside of Dr Alham’s lectures, the course lacked direction and the topics discussed were not as relevant to modern/advanced NLP. I have a feeling that many lectures were wasted on topics that could be a quick google search. Worst of all, we did not receive ANY feedback or grades throughout the semester (except for assignment 2 which is managed by Dr Alham...) |
| NLP702 (2024) | Suggest to Dr. Alham Fikri Aji to take bigger part of the course |
| NLP702 (2025) | A lot of professors struggle to say "i don’t know" when asked a question they don’t have the answer to. Dr. Alham is one of very few professors that is comfortable and confident saying he doesn’t know instead of answering something vague and it has been incredibly refreshing, especially given how technically strong he is. As someone with massive impostor syndrome, this normalization of not knowing everything is really important and helpful and I find myself less likely to try to cover my own tracks with dodgy answers when someone asks me a question I can’t answer. |
| NLP702 (2025) | Dr. Alham is the best! He can deliver complex concepts in a way that students understand. The pace of teaching is good. Funny. It would be better if he takes all parts of NLP702 next year :) |
| NLP702 (2025) | he could explain everything in newbie terms, so i could understand easily |
Advising and Mentoring
My group organization is primarily flat. Students, postdocs, and assistants can easily come to my office or ping me via Slack or Discord to discuss any topic. Likewise, collaboration is encouraged; I urge them to discuss ideas with one another rather than working in isolation. Students frequently drop by my office to discuss research or simply to chat. This structure ensures I remain involved in all research projects and hands-on with many of them. I strive to avoid a purely managerial role where the research details are obscured, or letting the group become so hierarchical that I lose touch with my students’ work.
To date, I have advised 5 PhD students, 9 MSc students, and several research assistants and visitors. I have supervised one postdoc, and another is joining soon. Within the MBZUAI NLP department, this represents a significant group size, particularly for an Assistant Professor.
Beyond MBZUAI, I actively collaborate with and advise external students, primarily BSc and MSc students from Universitas Indonesia and ITB, totaling 15 students so far. Several of these students have continued their studies at MBZUAI; I view this as a valuable investment in the talent pipeline that also allows me to contribute back to my country. While maintaining a flat structure, I encourage PhD candidates to mentor master’s or undergraduate students to develop their own mentorship skills, though I remain actively involved in these projects as well.
My expectations for PhD students extend beyond the completion of a thesis. My goal is to mold them into exceptional researchers equipped to secure strong positions after graduation. Consequently, my mentorship goes beyond technical supervision. We frequently discuss career strategy and professional growth to ensure they are fully prepared for the field.
I do not demand a specific number of publications, though I do expect students to publish to avoid the “red flag” of a zero-publication track record. However, publication count is not the main focus. In an era where tens of thousands of papers are submitted each cycle, adding one more unrecognized paper is of limited value. Instead, I ask my students to aim for influential and impactful work, even if it is only one single paper.
While impact is harder to quantify, I define success through indicators such as community discussion, dataset or model adoption (downloads/usage), and citations by other influential works. I believe that this approach is far more valuable than accumulating dozens of unrecognized publications in a sea of papers.
I recognize the critical importance of internships. Securing an internship at a top tech company or reputable research lab is a mandatory requirement for the members of my group. I actively assist students by providing internal references to tech companies and connecting them with colleagues at various institutes. In addition, I support them with CV review and interview preparation when necessary.
Future Teaching and Mentoring Plan
One of the challenges of teaching at a relatively new university is the limited number of students. Nevertheless, I see MBZUAI growing rapidly and having more students in the near future, so I am very happy to advise more PhD and MSc students from MBZUAI.
In terms of teaching, I will continue to teach NLP702/NLP806 at MBZUAI. MBZUAI has also grown as a university and now we have our first cohort of undergrads of around 100 students. The university trusts me to lead teaching this first cohort on Algorithms and Data Structures and Algorithms in 2026.
Beyond that, I am proactively building an undergraduate team to compete in the ICPC, a highly prestigious international university-level competition in competitive programming. I recruited 15 top students and organized several selection contests. I now have 2 teams, each consisting of 3 students, and I am coaching them for the ICPC regional contest this coming December. Although ICPC training and DSA are not closely related to NLP, this work has been especially rewarding because it reconnects me with my long-standing passion for algorithms and coding.