Teaching
Teaching
Making machine learning and AI intuitive, rigorous, and applied, for students at LUMS and for thousands of professionals and learners beyond it.
Intuitive, Rigorous, and Applied.
My aim as a teacher is to ignite a genuine hunger for knowledge: to expose the structure, elegance, and relevance of a problem so that students build an intuitive, holistic understanding before a single equation appears. I never write an equation without explaining what goes on behind the symbols, I anchor ideas in locally relevant examples (often interspersing Urdu), and I lean on live demos, animations, and thought experiments. In my Speech Processing course I open by projecting a live spectrogram of my own voice; in Machine Learning I explain bagging through the wisdom of a crowd guessing jellybeans before naming a single algorithm. Assessment, in my courses, is steady practice rather than high-stakes drama: weekly quizzes, frequent programming assignments, and reading assignments on ethics, fairness, and the societal impact of AI.
Transforming machine learning at LUMS
When I arrived at LUMS, the graduate Machine Learning course (CS-535 / EE-514) was an under-enrolled elective; its last offering before me had 20 students. I rebuilt it from scratch around the "Intuitive, Rigorous, Applied" mantra, lowered the entry bar with supplementary background videos, and opened it across disciplines to anyone with introductory statistics, linear algebra, and programming. Enrollment rose to a record 208 students, with strong representation from non-CS and engineering majors, and consistently high evaluations (instructor and course means around 4.5 to 4.6 out of 5). Many students call it the best course they have taken at LUMS. My graduate Foundations of Generative AI (CS 5302) and Speech Processing courses have seen the same enthusiastic response.
Demystifying generative AI
Recognizing how urgently generative AI needs to be made legible to non-specialists, I have, since February 2023, led more than 85 training programs reaching over 3,000 professionals: C-suite and executive leaders, vice-chancellors, deans and directors, 300 doctors, 300 faculty and staff, civil servants, and PhD students. Delivered directly and through the LUMS Learning Institute and the Pakistan Society for Training and Development, these programs have served organizations including Mercy Health (USA), Jazz, Engro, Pakistan State Oil, OGDCL, Faysal Bank, SNGPL, and The Citizen's Foundation. They cover AI's affordances, limitations, and ethics, and the practical strategies that help people build accurate mental models and integrate AI into their work.
Open courseware
I believe in democratizing knowledge. My full courses are freely available on YouTube and have been watched more than 175,000 times, with notes of thanks from learners around the world. With LUMSx, the university's distance-learning arm, I developed an asynchronous Machine Learning course that uses mixed English and Urdu instruction and locally relevant examples to reach learners who face a language barrier with existing resources.
Machine Learning
A complete graduate course, intuitive, rigorous, and applied, watched 175,000+ times.
Speech Processing
Foundations of speech recognition and synthesis, taught in mixed English and Urdu.
Generative AI (Advanced NLP)
The principles and practice of modern generative models and large language models.
Fundamentals of Computer Systems
A ground-up introduction to how computers and computation actually work.
Browse the whole channel on YouTube →
Recognition
Vice Chancellor's Award for Teaching Excellence, LUMS (2023–24), the university's highest teaching honor. My teaching spans LUMS, ITU, and FAST-NU, and includes teaching as a doctoral student at Carnegie Mellon.
For student supervision and placements, see Students & mentorship.