CV
The full CV is a PDF.
Positions
- Sep 2026
to presentResearch Assistant, University of Massachusetts Amherst
Amherst, MA
- Estimating blood pressure from graphene electronic tattoo bioimpedance with physics-informed and tree-based models, with the Kireev Lab.
- Extending the Bosch internship project with Bosch Research by reconstructing the program execution traces of industrial controllers from process sensors (IAAI-27, under review). The method detects code-path faults that sensor readings alone cannot reveal at 0.99 AUC.
- Placed 10th of 42 teams in the PhysioNet Challenge 2026 with interpretable sleep features for predicting future cognitive impairment (0.678 age-conditioned AUROC on the hidden test set; CinC 2026, accepted).
- May 2025
to Aug 2025AI for Cyber-Physical Systems Intern, Robert Bosch LLC, Bosch Research
Pittsburgh, PA
- Built a contrastive model aligning vision, sensor, program trace and text data from an industrial control testbed for language-guided few-shot classification of machine state.
- Aug 2023
to Aug 2026Research and Teaching Assistant, Worcester Polytechnic Institute
Worcester, MA
- Developed LLaSA, a sensor-aware LLM that answers questions about human activity from IMU data and outperforms GPT-3.5 and GPT-4o-mini, with 235,661 released training pairs (SensorCap and OpenSQA).
- Developed JOLT, an early-exit training method that ranked first of 11 methods across 7 datasets under compute budgets.
- Served as teaching assistant for the graduate course On-Device Deep Learning and three undergraduate ECE courses.
- Feb 2021
to Jul 2023Machine Learning Engineer, Hishab Technologies
Dhaka, Bangladesh
- Built telephony conversational AI for Bengali speakers, covering speech recognition, text to speech, named entity recognition, language models and grapheme-to-phoneme conversion.
- Deployed services for banking and microfinance clients, including Bank Asia and Hitachi India.
Education
- May 2028 (expected)
Ph.D., Electrical and Computer Engineering
University of Massachusetts Amherst
Advisor Bashima Islam. Began at Worcester Polytechnic Institute in Aug 2023; transferred with the lab in 2026.
- Dec 2025
M.S., Electrical and Computer Engineering
Worcester Polytechnic Institute
GPA 4.00/4.00. Thesis: Interpretable and Efficient AI for Sensor-Rich Mobile and Embedded Systems.
- Feb 2021
B.Sc., Electrical and Electronic Engineering
Bangladesh University of Engineering and Technology
Thesis: Audiovisual emotion recognition using multi-head attention based neural network, advised by Celia Shahnaz.
Awards
- 2019, 20202nd place (ICIP 2019) and 5th place (ICIP 2020), IEEE Video and Image Processing Cup
- 20204th place, IEEE Signal Processing Cup at ICASSP 2020
- 2018National champion and 22nd worldwide, IEEEXtreme 12.0 programming contest (also national champion in 2017)
Service
- 2019 – 2021
Chair, IEEE Signal Processing Society Student Chapter, BUET
- Organized the 2019 IEEE SPS Winter School on Multimodal Signal Processing.
- Grew it into the largest IEEE student branch chapter in Bangladesh, earning an IEEE SPS growth reward.
- 2017 – 2019
Ambassador and Country Lead (2019), IEEEXtreme Programming Contest
Skills
- ProgrammingPython, PyTorch, TensorFlow, Keras, scikit-learn, MATLAB, C/C++, Git, LaTeX
- Machine learningMultimodal and contrastive learning, LLM fine-tuning, on-device inference, physics-informed models
- SignalsBiomedical signals (PPG, bioimpedance, sleep PSG), IMU, audio and speech, image and video
- SystemsSLURM clusters, MLOps, GCP and AWS, Kafka, Jenkins, microcontrollers, FPGA