I am currently a summer research intern with Microsoft Research, New England. I am working in the social media collective (https://socialmediacollective.org/) with danah boyd. We are studying emergent efforts in the disaster planning & response community to consider population level and socio-demographic factors in risk modeling.

I am supported by an NSF Grant to study technology in fire-based EMS and the changing nature of risk work. I am particularly focused on dispatch systems. (https://risk-it.github.io/)

PI: Professor Joshua BlumenstockSupported a team partnering with the government of Togo to develop, advertise, and evaluate a mobile cash transfer program designed to slow the spread and mitigate the economic effects of Covid-19. • Designed a novel Algorithm to predict location from mobile phone data in sparsely populated areas (Standalone Paper forthcoming).• Developed a model to predict occupation from mobile phone data and completed an analysis of mobility and income and behavior changes for taxi drivers during the Coronavirus crisis for the Ministry of Transportation.• Completed a causal analysis of the efficacy of text messages to drive program enrollment.

PI: Dierdre Mulligan (Spring 2020) Researching how large tech companies operationalize definitions of fairness and ethics in machine learning through document analysis and interviews.

Recipient of year long fellowship through both the Center for Technology Society and Policy and The Center for Long Term Cyber Security. The project is called "The public defense project", and we are collaborating with Secure Justice, an Oakland non-profit to help bridge the skills gap between prosecutors and public defenders. In particular are completing qualitative research to understand public defenders experience with cellphone extraction data and location data in the courtroom, which will inform the development of a software tool to assist defenders in fairly representing their clients. Our won Sarukkai award for the Information Science masters capstone with the highest social impact.

Software engineer and data scientist for for Einstein Platform, a general machine learning platform which several machine learning apps at Salesforce including Einstein Prediction BuilderWorked with customers to identify problems with models by exploring data and running experiments using Python, Spark, SQL on AWS. - Designed pipeline for real time scoring metrics passing through 5 services - Collaborated on a cross-functional team of designers, lawyers, and data scientists at Salesforce to design data science features to allow customer design systems to surface and guard against issues of algorithmic bias in Salesforce’s machine learning suite- Spearheaded necessary Spark upgrade for Einstein Engagement Scoring pipeline which now serves more than a billion predictions per day.- Drove adoption of new experimentation framework among 20 person data science team by improving performance and usability.

I work as a software engineer on our graphical analytics platform, which is built in Scala, Java, and Javascript and help build out our machine learning capabilities mostly with Spark.Accomplishments- Worked to develop an algorithm which optimizes Spark submission settings based on cluster size, and utilization, input data size and algorithm requirements.- Helped design, and am now on a two person team maintaining, Alpine's customer-facing extensibility framework. The framework enables customers to add their own Spark or database algorithms to our visual analytics engine. In particular, I:- - wrote much of our parsing engine for reading and writing files on HDFS including integrating support for Avro and Parquet- - added date time support for Hadoop- - added the ability to produce data visualization and well formatted tables in the framework- Co-Authored Alpine's first NLP package with Spark. We added support for stop-word removal, sentence tokenization, stemming, n-gram parsing, TF-IDF, and LDA with Spark.- Added support for Collaborative Filtering on Hadoop. Our implementation is based on that in Spark MLlib, but also includes support for string keys and ranking predictions for users.- Worked to rapidly develop customer-facing data processing routines at scale including, calculating rank statistics, custom aggregations on wide data, and computing inferential statistics.- Lead formal trainings for customers from organizations around the world on using our extensibility framework. Also trained the members of our data science team (who had no pervious Spark experience) to implement their own algorithms in our extensibility framework.

Full time assistant for three courses1. Software Engineering: a project based class, taught Javascript, with angular JS and testing frameworks, to help student teams develop end to end web applications)2. Intro to Computer Science (Java and Robot C)3. Algorithms and Design: Taught as an applied version of a similar course offered at StanfordAccomplishments: - Fulfilled my responsibility to design rubrics and oversee grading for weekly assignments in the intro class- Ran well attended review sessions- Created a series of handouts for difficult algorithms topics to be used next year- All groups in software engineering class completed two working web applications

Appointed chair of a committee made up of the VP of finance, faculty, alumni, and students that successfully created significant policy changes in how Wesleyan both manages their operating account and pursues shareholder advocacy with their endowment.I Drafted share-holder resolutions, coordinated meetings and organized events and open forums and helped support related student groups.Accomplishments: - While a member we pushed the university to move $500,000 of general operating account into local banks. As chair I spearheaded a successful effort to double that number.- Organized our proxy voting system, resulting in significant increase in shareholder participation.

As a continuation of my work for the Wesleyan media project, I built an application to help us better collect data for my senior capstone project in computer science under the aegis of Computer Science Professor Norman Danner. The application scrapes articles from an academic database, uses a regular expressions algorithm to parse the articles into text attribute strings, and then uses a machine learning algorithm I developed using an ML tool called WEKA, to discard the articles that are not relevant to the analysis. Accomplishments: Ultimately, the application reduced the number of articles which had to be scored by hand by 20%.

Selected for the paid Quantitative Analysis Summer Apprentice Program, included daily instruction in statistics and data mining, and analysis tasks for a Government Professor, Erica Fowler. My independent project examined the relation between print media coverage of campaign advertisements and campaign strategy using survival analysis and logistic regressions implemented in R, Stata, and Python. Accomplishments: At the end of the summer, my research partner and I received funding to continue our research, which we presented at an academic conference in 2013.

Used Hive on Hadoop cluster to mine consumer data and the Python, R, to process that data and present accessible, quantitative insights to support marketing and sales designs and product features.In particular, I Focused on analyzing differences across modes of usage such as subscription vs. free usage, and mobile vs. automotive users.Accomplishments: At least one product feature implemented from my recommendations.

I worked as a professional white water rafting guide on the pigeon river, leading several trips a day down class four rapids.