Matt Rossman Email & Phone Number

Worcester, Ma, Us

Building Hallway's web platform for avatar identity. Responsible for the design and development of our avatar marketplace and user profiles. Collaborating with artists to distribute optimized, high quality 3D assets. Talking with users to validate our decisions.

GRA in the Augmented Environments Lab, integrating productivity tools in Hubs Cloud to support remote coworking in social VR.

St Louis, Missouri, United States

Developed client-facing demos of new features for Mozilla Hubs (open source web-based social VR platform) including visual effects and a voice-enabled conversational AI bot. I also overhauled the internal scene editor with a configurable component system, improving the team's code re-usability and scene maintainability.

CS 397A Predictive Analytics with PythonIn my first TA position, I helped a class of 40 students without much or any CS background grow comfortable with programming in Python and the PyData stack. I spent 1-on-1 time with students in my office hours, made 100+ contributions on the class Piazza forum, and left detailed feedback on assignments. My goal was to provide students with intuitive and straightforward explanations to get them past the hurdles I experienced back when I was learning Python and the command line.

Arlington, Tx

As people age, falls can become a debilitating risk. What if caretakers could visually monitor an elderly patient's gait without actually being there, and without intrusive cameras that strip away privacy? This summer, my team proposed and prototyped a solution for ambient gait monitoring through 3D skeletal reconstruction, inferred from pressure patterns on a low cost smart floor.During this REU (Research Experience for Undergraduates), I worked with Python data analysis libraries to infer gait from raw, low resolution pressure readings. I collected a training set of synchronized Microsoft Kinect skeletal recordings and developed a model to infer skeletal motion from pressure data. These efforts turned into a research paper that was published in PETRA 2020.Tools: Xarray, Kinect v2, Plot.ly, Three.js

Boston

THE WHAT:I took a big, bloated, broken testing suite and made it faster to run and easier to extend.THE HOW:I started with the old testing framework which had grown to accommodate all of the product's domains. I salvaged the parts that were relevant to my team's domain (membership). I then redesigned the abstractions to fit in with the team's API patterns.To speed up testing, I changed the pipeline to populate and containerize the test database only once after the test suite is updated, rather than running this long process every single time the tests need to be executed.THE TOOLS:Docker, Spring Boot, Maven, Groovy, Jenkins

Andover, Ma

THE WHAT:I developed a Python module for forecasting building resource usage (e.g. energy, water) based on sensor data and contextual information. This was to be used by the town of Andover, MA for detecting anomalous energy usage in their municipal buildings.THE HOW:I read lots of research papers on various machine learning methods for trend forecasting. I took their findings to build a random-forest model that best applies to the data logs I was working with. I used proven statistical methods to flag behavior as anomalous against predictions. I blogged the entire process.THE TOOLS:Anaconda, Pandas, Numpy, Scikit-learn, SciPy

Master Of Science - Ms, Human Computer Interaction

Bachelor Of Science (B.S.), Computer Science, 3.96

Activities and Societies: UMass Outing Club, Juggling Club, ACMCommonwealth Honors College

Study Abroad, Ingenería Informatica

Madrid, Spain Spring '19

Bachelor Of Science (B.S.), Computer Science, 4.0

Attended for one semester before heading to the UMass Amherst Honors College. Relevant Courses: Discrete Structures, Accelerated.