Career Profile

I am a Ph.D. candidate at the Thayer School of Engineering, Dartmouth College. My research interests are electronics and machine learning with a focus on using learning algorithms and signal processing tools to implement and investigate wearable health monitoring devices for patients with lung and airway diseases.

Experiences

Research Scientist Intern

Summer 2017
Amazon Inc., Alexa Machine Learning, Cambridge, MA

Performed speech recognition research at Amazon for the Alexa conversational AI system and platform. Work included various machine learning tasks pertaining to Acoustic Event Detection (AED), Voice Activity Detection (VAD) and Natural Language Understanding (NLU).

Postbac Research Assistant

2013 - 2014
Analog Electronics Lab, Thayer School, NH

Investigated the feasibility of a piezoelectric-based sensor as wearable device for monitoring asthma symptoms under Dr. Kofi Odame’s supervision. I designed and implemented the piezo-sensor electronics. I explored features of interest in the sensor output and put together signal processing methods to extract the said features. Finally, I explored different machine learning algorithms to automatically detect coughs in the sensor output stream. In this project, I garnered skills in machine learning algorithms, acoustic signal processing and printed circuit board design.

Postbac Research Assistant

2013 - 2014
Clinical Applications of Bioimpedance Lab, Thayer School, NH

Performed research with Dr. Ryan Halter on a small form-factor bioimpedance system. I devised a portable tetra-polar impedance analysis system to meet higher efficiency and safety specifications. This involved developing an analog front-end circuit to adapt a stock impedance analyzer chip to meet the higher requirements. In this project, I gained experience in circuit design and implementation for clinical electronic instruments.

Undergraduate Research Intern

Summer 2012
Molecular Imaging Instrumentation Lab, Stanford University, CA

Conducted research in the MII lab of Dr. Craig Levin, in Stanford University’s Radiology Department. We designed a Cadmium Zinc Telluride (CZT) based PET system for imaging small animals. Specifically, I calibrated the detector unit to yield energy spectra of event pulses in units of kev versus event counts. I also devised a system to automatically determine a threshold for noise and trigger the detector on that threshold. From this experience, I acquired skills in digital signal processing and object oriented programming in C++.

Undergraduate Research Assistant

2011 - 2013
Computational Plasma Physics Lab, Dartmouth College, NH

Conducted research with Dr. Richard Denton in investigating the dependence of bulk plasma composition at geosynchronous orbit on solar cycle phases. Using MATLAB, we investigated and modeled the changes in particle density of bulk ions with respect to solar cycles, magnetic local time, and seasons. I acquired computational and analytical techniques required to efficiently manage and process enormous data.

Projects

Find below some projects I have worked on in the past and their descriptions. Feel free to contact me for more information on any of these.

DeepCough - Deep neural network architectures for detecting cough sounds from a chest worn acoustic sensor.
Acoustic Health Sensors - Circuitry and machine learning algorithms for analyzing respiratory acoustics for detection and characterization of Asthma or COPD symptoms.
Portable BioImpedance Device - An embedded system extending a single chip biopolar impedance analyzer IC to a tetrapolar device for applications in analyzing cancerous tissues during prostate biopsies.
MHB App - Methodist Hymn Book Android app for the Methodist Church, Ghana. An collation and indexing of all hymns, creeds, and service routines as available in the physical methodist hymn book used in all methodist churches in Ghana.
Timely App - An android app for Dartmouth students that anticipates what information user's need at any particular point in time, and delivers it to them in a Google-Now-cards kinda approach. Information includes current meal options at dining halls, events in buildings user might be in close proximity with, and so on.

Skills & Proficiency

Machine Learning - Python, Tensorflow & Pytorch

Embedded - C/C++, Unix/Bash, ARM & Arduino

Electronics - SPICE, VHDL & PCB Design

Signal Processing - MATLAB

Design - SolidWorks, Sketch & Photoshop

Others - Java, Android & Mathematica