What I work on

I’m interested in intelligent sensing systems that work outside the lab: embedded hardware, efficient on-device AI, and the real-world data they produce, especially for health and agriculture.

Since 2023 I’ve been a research assistant at the Microsystems and Nanoengineering Lab. Most of my work there sits at the intersection of hardware, sensing, and machine learning: building the device, writing its firmware, training the model, and then calibrating the pipeline with feedback from field trials.

MAST-CloudNet started as an attempt to tackle dengue, one of the biggest public health problems in Bangladesh. The idea was to spot emerging Aedes hotspots early, so that authorities could act before an outbreak took place. The system lures male Aedes with a sound that mimics female wingbeats, photographs the mosquitoes, identifies the species with deep learning and sends live statistics to a web dashboard. I first tried inference on the Raspberry Pi itself with smaller models and quantization. Although this approach kept the end-to-end latency down, it was not accurate enough to tell species apart, so I moved inference to a cloud server and kept the Pi for sensing and capture. That trade-off on the Pi is what got me interested in efficient AI: making models small and fast enough for cheap hardware without losing the accuracy that matters.

Before that, I built a two-axis CNC camera rig and trained a deep learning model to track seed germination for a nanopriming study in agriculture.

You can find more on my publications and projects pages.

Outside the lab

I spend my free time playing guitar, writing automation scripts, learning Zig (this site is built with Zine) and reading classics. Here is a list of the books still on my shelf, thanks to my friends.

Meet Bono

Say hi to my friend Yassine Bono!