Hello! I'm Mathew.
I'm a data scientist, engineer, and aspiring software developer.
About Me
I love to build things - that's why I became an engineer! Along the way, I fell in love with electronics and programming, despite classically training as a Chemical Engineer, because it was so easy to rapidly iterate on ideas and see them come to life in real-time. Throughout my 9+ years in industry I've been fortunate enough to partner with some amazing people in supply chain, marketing, E-Commerce, and payments! When I'm not working as an analytics manager at Visa, I'm building things, traveling, gardening, and learning a new instrument!
'%3e%3cpath%20d='M54.158%2051.1466H56.6514C56.0792%2052.5384%2055.2378%2053.5715%2054.2208%2054.298C52.9111%2055.2336%2051.2381%2055.7112%2049.3175%2055.7112C46.6291%2055.7112%2044.5144%2054.7164%2043.0603%2053.0578C41.5942%2051.3855%2040.7313%2048.9577%2040.7313%2045.9958C40.7313%2040.3132%2044.2952%2036.2411%2049.3175%2036.2411C52.1695%2036.2411%2054.2397%2037.5199%2055.5905%2039.5025C56.8183%2041.3046%2057.4696%2043.7243%2057.4725%2046.3354H44.5572H43.512L43.5582%2047.3796C43.6472%2049.3903%2044.2193%2051.0111%2045.2996%2052.1337C46.3905%2053.2673%2047.8746%2053.7635%2049.5145%2053.7635C51.7063%2053.7635%2053.4032%2052.6016%2054.158%2051.1466ZM43.5937%2043.7544L43.5002%2044.8402H44.59H53.6017H54.73L54.5945%2043.7201C54.4136%2042.2249%2054.0197%2040.8084%2053.1491%2039.7569C52.235%2038.653%2050.9106%2038.0855%2049.1862%2038.0855C47.4766%2038.0855%2046.1055%2038.7143%2045.1431%2039.7982C44.2025%2040.8577%2043.7214%2042.2718%2043.5937%2043.7544ZM114.285%2051.1466H116.779C116.206%2052.5384%20115.365%2053.5715%20114.348%2054.298C113.038%2055.2336%20111.365%2055.7112%20109.445%2055.7112C106.747%2055.7112%20104.633%2054.716%20103.182%2053.0583C101.718%2051.3865%20100.859%2048.9588%20100.859%2045.9958C100.859%2040.3132%20104.422%2036.2411%20109.445%2036.2411C112.297%2036.2411%20114.367%2037.5199%20115.718%2039.5025C116.946%2041.3046%20117.597%2043.7243%20117.6%2046.3354H104.684H103.644L103.685%2047.3752C103.765%2049.3889%20104.338%2051.0107%20105.419%2052.1337C106.509%2053.2668%20107.993%2053.7635%20109.625%2053.7635C111.829%2053.7635%20113.53%2052.6029%20114.285%2051.1466ZM103.705%2043.7483L103.604%2044.8402H104.701H113.729H114.857L114.722%2043.7201C114.541%2042.225%20114.147%2040.8077%20113.273%2039.7557C112.356%2038.6517%20111.028%2038.0855%20109.297%2038.0855C107.594%2038.0855%20106.227%2038.7157%20105.266%2039.7982C104.327%2040.8559%20103.842%2042.2677%20103.705%2043.7483ZM20.5917%2043.0255H19.5917V44.0255V55.159H16.4374V28.9781H19.5917V39.5166V40.5166H20.5917H31.7209H32.7209V39.5166V28.9781H35.8751V55.159H32.7209V44.0255V43.0255H31.7209H20.5917ZM77.6014%2043.6318V55.159H74.7591V44.3141C74.7591%2042.5678%2074.4832%2041.0726%2073.7426%2039.9923C72.9505%2038.8367%2071.7396%2038.3117%2070.2791%2038.3117C68.6655%2038.3117%2067.2823%2038.8159%2066.3562%2040.087C65.4826%2041.286%2065.1426%2043.005%2065.1426%2045.1424V55.159H62.2839V36.7934H64.88V38.6294V39.6294H65.88H65.9949H66.577L66.8645%2039.1232C67.9722%2037.1724%2069.893%2036.2411%2071.7236%2036.2411C74.2572%2036.2411%2075.6171%2036.9985%2076.4067%2038.1211C77.2531%2039.3247%2077.6014%2041.1568%2077.6014%2043.6318ZM97.8094%2036.7934L92.0939%2044.1156L91.6423%2044.6942L92.0589%2045.2985L98.8572%2055.159H95.2803L88.6045%2044.8713L94.5844%2036.7934H97.8094ZM86.5803%2045.13V47.6057V55.159H83.7379V28.9781H86.5803V41.7518V44.5742V45.13ZM70.8372%201.57933C90.1617%201.57933%20107.618%206.32294%20120.21%2013.9358C132.822%2021.5602%20140.404%2031.953%20140.404%2043.2398C140.404%2054.5285%20132.822%2064.9204%20120.211%2072.5434C107.618%2080.1549%2090.1618%2084.8968%2070.8372%2084.8968C51.521%2084.8968%2034.0686%2080.155%2021.4784%2072.5434C8.86912%2064.9204%201.28662%2054.5285%201.28662%2043.2398C1.28662%2031.953%208.86912%2021.5601%2021.4784%2013.9358C34.0687%206.32291%2051.521%201.57933%2070.8372%201.57933ZM70.8372%203.42728C52.2095%203.42728%2035.283%207.81492%2022.9792%2014.9472C10.7018%2022.0641%202.84862%2032.0374%202.84862%2043.2398C2.84862%2054.447%2010.7021%2064.4123%2022.9796%2071.5201C35.2836%2078.6431%2052.21%2083.0204%2070.8372%2083.0204C89.4645%2083.0204%20106.391%2078.6431%20118.695%2071.5201C130.972%2064.4123%20138.826%2054.447%20138.826%2043.2398C138.826%2032.0374%20130.973%2022.0641%20118.695%2014.9472C106.391%207.81492%2089.4649%203.42728%2070.8372%203.42728ZM125.253%2028.9781V55.159H122.411V28.9781H125.253Z'%20stroke='url(%23paint0_linear_2003_272)'%20stroke-width='2'/%3e%3c/g%3e%3cdefs%3e%3clinearGradient%20id='paint0_linear_2003_272'%20x1='141.404'%20y1='43.2381'%20x2='0.286621'%20y2='43.2381'%20gradientUnits='userSpaceOnUse'%3e%3cstop%20stop-color='%239E00FF'/%3e%3cstop%20offset='1'%20stop-color='%23FB8A05'/%3e%3c/linearGradient%3e%3cclipPath%20id='clip0_2003_272'%3e%3crect%20width='142'%20height='86'%20fill='white'/%3e%3c/clipPath%3e%3c/defs%3e%3c/svg%3e)
Projects
NLP/LLMs | Improving ELECTRA Performance
Pre-trained models that over-rely on popular annotated datasets, like SNLI and MNLI, can learn dataset specific “Artifacts”. This hinders model generalization and leads to poor performance when deployed to production.
Deep Learning | Teaching AI to Drive
CNN (Convolutional Neural Networks) are a type of feed-forward deep neural network that leverages convolution layers for visual processing. In this project, I leverage a low-level ‘auto-pilot’ controller to train an AI vision based driving system to complete a course of SuperTuxKart.
Devops | A Modern Data Analytics Stack In a Box
“Big Data” no longer requires a complicated cluster of machines and a PhD. We can use Docker to run a local analytics stack! I’ve used this stack to train teams at Amazon/WFM on the internals of PySpark and Docker.
App Development | Business Friendly UI/UX
Most business stakeholders err away from using CLIs (Command Line Interfaces) or from writing complicated SQL. Instead, I designed and built an application that accepts input, queries a MS SQL Server Database, and returns results for master data management teams to maintain tables in SAP.
Forecasting | Using SARIMA to Predict Dallas Weather
SARIMA (Seasonal Auto Regressive Integrated Moving Average) is a widely used model that brings components of different Time Series forecasting methods to create a more robust way to forecast. We can use SARIMA to predict the “one-year-in-advance” prediction of temperature for January 2018.
Kalman Filters | Univariate Dynamic Linear Model Time Series
Kalman Filters have many applications in aerospace, navigation, econometrics, etc. and are often applied in time series analysis for signal processing. We are able to estimate a joint probability distribution over the variables by taking the uncertainty into account.