Electrical engineering · Machine learning · Energy systems
Research engineer working on machine learning for energy systems, air quality and transport behaviour. Six and a half years in applied engineering, four of them in a dedicated research role. Seeking a funded doctoral position.
Applied engineering
6.5
years
4 years in a dedicated research role
Peer-reviewed output
4
papers
+1 in preparation · 2 theses
MSc ICT · CGPA
3.81
/ 4.00
Earned in full-time employment
One question, carried
14
years
2012 thesis → ML extension today
Profile
I am an electrical and electronic engineer with six and a half years in applied engineering, four of them in a dedicated research engineering role, and an MSc in Information and Communication Technology completed alongside full-time work at Bangladesh University of Professionals.
My published work covers applied machine learning on environmental and public-health data, model interpretability, and health technology adoption. My longest-running line of work runs from my 2012 undergraduate thesis to its machine-learning extension today, and asks a single question: how should a supply-constrained national grid be scheduled so that load shedding falls?
Since 2019 I have also operated large-scale algorithmic advertising systems. That is not a detour. It is sustained practice in controlled testing at population scale, and a view of automated decision-making from inside the system rather than from the outside.
Doctoral directions
Each direction below is already load-bearing somewhere in my record — a thesis, a paper, or four years of operating the systems in question. I am open to a supervisor reshaping any of them.
Does switching vehicle type change how the same individual drives? My MSc thesis built deep-learning detection of over-speeding and lane changes from naturalistic footage of mixed, weakly regulated urban traffic — a setting almost absent from a literature trained on regulated European and North American roads.
Including the comparison between rationing by outage and rationing by connection restriction. This is the decade-long thread: from peak-shaving hydro dispatch in my BSc thesis to learned dispatch policies under chronic generation shortfall today.
With particular attention to tropical siting and the Johor–Singapore cluster. It joins my grid-side training to the environmental modelling in my 2026 conference work, and sits on my doorstep.
For high-cost, infrequent purchases such as vehicles — and, relatedly, how optimisation systems reach targeting decisions, and where regulatory assumptions diverge from operational reality. I have run these systems daily since 2019.
Supervised machine learning, model benchmarking and validation · explainable AI (LIME, SHAP, feature importance) · deep learning for computer vision · experimental design and controlled A/B testing at population scale · survey design · in-depth interviewing and thematic analysis · research ethics procedure and institutional approval · Python, including deep-learning implementation.
Publications
ORCID 0009-0009-8360-9690. Author name in bold. Full texts on request.
7th Multimedia University Engineering Conference (MECON2026 / DIFCon 2026), Multimedia University, Malaysia, 5–7 May 2026
A predictive framework built on open-source meteorological data to forecast dengue outbreak trends across Malaysia, aimed at early warning and public-health preparedness. Affiliations: Multimedia University, Malaysia; Miraiyo, Bangladesh; Centre for Electric Energy and High Voltage Engineering.
7th Multimedia University Engineering Conference (MECON2026 / DIFCon 2026), Multimedia University, Malaysia, 5–7 May 2026
Benchmarks machine-learning algorithms against seasonal operational features of brick kilns to forecast severe air-quality degradation in Dhaka — a source term that generic pollution models routinely omit.
International Journal of Advanced Research in Computer and Communication Engineering (IJARCCE), March 2024
Compares LIME, SHAP and conventional feature importance on a decision-tree classifier. Finds high local precision and recall for LIME, a local–global balance for SHAP, and sets out practical selection criteria for interpretability frameworks.
Asian Journal of Multidisciplinary Research & Review (AJMRR), 2(2), 592–621, 2021. ISSN 2582-8088
Mixed-methods study of how AI-driven telemedicine was implemented and adopted in Bangladesh during the pandemic. Thirty-five in-depth interviews with practitioners, patients and platform staff; purposive sampling; instrument designed and all data collected by the author under institutional ethics approval.
In preparation
Replaces the rule-based peak-shaving schedules of my undergraduate thesis with learned dispatch policies under chronic generation shortfall. The same question, fourteen years on, with the methods that were not available the first time.
Bangladesh University of Professionals, Faculty of Science and Technology, Dhaka. Examined March 2022
Deep-learning models applied to naturalistic traffic footage to detect over-speeding and lane-change manoeuvres in mixed, weakly regulated urban traffic — a setting under-represented in driver-behaviour research.
Ahsanullah University of Science and Technology, Dhaka, 2012
Dispatch scheduling to flatten demand peaks in a chronically supply-constrained national grid. Extended a decade later with machine-learning methods — see P5.
Where I have been
Research collaboration — environmental and climate-health machine learning
Multimedia University, Malaysia · Miraiyo, Bangladesh · Centre for Electric Energy and High Voltage Engineering. Supervising collaborator: Dr Siow Chun Lim. Output: P1 and P2.
MSc, Information and Communication Technology — CGPA 3.81 / 4.00
Bangladesh University of Professionals, Dhaka. Completed in full-time employment. Group thesis on deep-learning speed and lane-change detection.
Research Engineer
Debonair Electrical & Electronics Ltd., Dhaka. Four years of applied product research and technical evaluation, reported to international suppliers and industrial clients; trained junior engineers.
BSc, Electrical and Electronic Engineering
Ahsanullah University of Science and Technology, Dhaka. Thesis on peak-shaving hydro dispatch and load shedding.
Social Media & Content Manager, Paid Media
Black Swan Business Setup Services, UAE (remote). Paid acquisition across five markets; builds and runs the full measurement stack (GA4, Tag Manager, Conversions API, Looker Studio). Population-scale controlled testing with measured outcomes.
Technical Writer · Marketing Manager · Assistant Manager
BuddyBoss, Canada · STUDiLOG · Enzaime Bangladesh. Developer documentation and pre-release testing; the Enzaime role provided the field access behind P4.
Support Engineer
Ahsan Automation Ltd., Dhaka. Industrial automation design and implementation; engineering drawing; collaboration with international manufacturers on new products.
Bengali — native, and a working research capability for fieldwork and document access in Bangladesh. English — full professional proficiency, medium of instruction throughout higher education. Working familiarity with Malay-language regional sources.
Contact
If any of the four directions above overlaps with your group’s work, I would value a short conversation. I can send a two-page research proposal, full texts of any paper listed here, and my transcripts on request.
I am a Bangladeshi national, currently resident in Malaysia, and available to relocate. My strongest independent evidence is publication P4 — a sole-authored, interview-based study run end to end under institutional ethics approval — and the fourteen-year thread from my undergraduate thesis to the dispatch work now in preparation.
The fastest way to reach me is email. I reply within a day.