Conditions in which every student can do the work well
The underrepresentation of women, minorities, and people with disabilities in computing is among the most consequential problems the field has. It matters to me as a teacher and researcher, and it matters to me personally.
Where this comes from
I was born and raised in Bangladesh and came to the United States for graduate study, which gave me a working understanding of what it takes for an academic environment to function for people arriving from different places. Five years at UMBC reinforced a simple observation that grounds my practice: differences in background coexist with a shared interest in learning and intellectual work. The teaching and mentoring choices I make follow from that.
In the classroom
I structure classrooms so that no assumption is made about a student on the basis of gender, ethnicity, religion, or prior preparation.
Active learning that accommodates varied preparation
I use live coding sessions and explain program logic statement by statement, so students can follow regardless of how much programming they arrive with. Teaching join operations in an introductory database course, I anchor relational concepts in structures students already know: a university enrollment system, a hospital records system. For students who need more grounding I hold extended office hours framed explicitly as low-pressure space for foundational questions, with priority for first-generation and underrepresented students.
Rotating leadership on team projects
Long-running team projects reproduce existing hierarchies when the same students always lead. I rotate team leadership through semester-long projects so every student, and particularly students from underrepresented groups, gets experience with project management and professional communication.
Inclusive design for online and hybrid formats
For asynchronous content I produce short modular videos rather than long single recordings, so students can revisit one technical segment at their own pace. Embedded quizzes and interactive polls sustain engagement in synchronous sessions. Anonymous mid-semester surveys arrive with enough time left to adjust pace, examples, and assessment format.
Through research
Algorithmic bias and unreliable inference impose their costs unevenly, and the heaviest costs land on communities that are already underrepresented or under-resourced. A clinical decision-support system that performs poorly on physiological signals from one population, a sea ice forecast uncalibrated for the regions most exposed to climate change, an intrusion-detection model that flags some groups disproportionately: each is a technical failure with a social consequence.
TimeGraph (KDD 2025), my open-source benchmark suite, is infrastructure for evaluating causal discovery methods before they reach sensitive domains. Technical rigor is a form of equity in this sense. Rigorous evaluation protects everyone, and it particularly protects the people who would otherwise absorb the cost of a method that performs well on average and fails in specific cases.
Institutional engagement
At UMBC I have engaged with the spirit of programs such as the Center for Women in Technology and the Meyerhoff Scholars Program, both nationally recognized models for supporting underrepresented students in STEM. As a faculty member I will seek out comparable programs at my home institution, advise undergraduate research with attention to recruiting students who would not otherwise picture themselves in a faculty member's lab, and support graduate mentorship structures that build community across cohorts.
What I want is to work in a department that learns from people of different backgrounds, and to give students of every background the conditions in which to do their work well: clarity about expectations, access to resources, structured support, and respect for the perspective each of them brings.