Mike's History

January 2017 - Present (2 Years, 4 Months)

Data Scientist, Big Data

San Francisco, CA, US

Talent Frameworks Visualization Module

Develop a Talent Frameworks Data Visualization tool for the Customer Success team. The visualization module produces the following visualizations: - Plot a histogram of all proficiency levels for a specific Skill ID. - Plot a pie chart of roles containing a specific Skill ID. The visualization is segmented by industry. - Plot a pie chart of roles containing a specific Skill ID. The visualization is segmented by Job Family. - Plot a histogram of importance levels for all skills within a specific Job Level Code. - Plot a histogram of importance levels for all skills within a specific Job Band Level and Framework Code. - Plot a histogram of proficiency levels for all skills within a specific Job Level Code. - Plot a histogram of proficiency levels for all skills within a specific Job Band Level and Framework Code.

Talent Frameworks Query Module

Develop a program for the Customer Succes Team that can run queries for a specific Skill ID in Talent Frameworks. We designed a module that can conduct the following Skill ID queries: - List all the roles that require a specific Skill ID. - Count the number of roles that contain a specific Skill ID. - Count the number of roles within each Job Family that contain a specific Skill ID. - Count the number of roles within each industry that contain a specific Skill ID. - Count the number of roles within each (industry, proficiency level) group that contain a specific Skill ID. - Count the number of roles for each proficiency level that contains a specific Skill ID.

Skill Score Algorithm

Implement the Skill Score Algorithm. Determine the optimal weightings of the individual components of the algorithm.

Talent Frameworks Job Band Segmentation

Develop a Job Band segmentation module for the Customer Success team. The module performs the following tasks: - List roles in a specific Job Band Level Code, which had expert and advanced skills associated with them. - Produce a Dataframe that presents a skill proficiency segmentation for a specific Job Band Level (level_code) and framework (fm_code).

Active Search Ranking Validation

Validate the Active Search Algorithm by calculating the Spearman Correlation Coefficient between the algorithm's rankings and the Recruiting Team's rankings of potential job candidates.

January 2016 - December 2016 (1 Year)

Data Scientist, Big Data

San Francisco, CA, US

• Real Estate Predictive Intelligence Client: Built a predictive model to determine the likelihood of a homeowner moving with a Random Forest classifier. • Energy Analytics Client: Using k- nearest neighbors, developed a time series disaggregation al- gorithm which deals with the disaggregation of individual energy appliances from the aggregate time series energy consumption data collected from a smart power meter. • Professional Talent Management Client (Talent Sky): Developed a search ranking algorithm which deter- mines the best candidates for a specific job query of desired skills. The algorithm takes into account the importance level and the proficiency level of the skills queried by applying a dy- namic statistical exponential decay penalty. The search ranking algorithm was prototyped in Python and deployed via ElasticSearch. Researching and developing recommendation systems for client’s professional talent management platform.

Search Match Score Algorithm

Developed a search ranking algorithm which determines the best candidates for a speci fic job query of desired skills. The algorithm takes into account the importance level and the pro ficiency level of the skills queried by applying a dynamic statistical exponential decay penalty. The search ranking algorithm was prototyped in Python and deployed via ElasticSearch.

Predictive Model Development for Real Estate Client

Built a predictive model to determine the likelihood of a homeowner moving with a Random Forest classi fier.

April 2015 - July 2015 (4 Months)

Data Science Fellow

San Francisco, CA, US

• Completed a 1000+ hour intensive data science fellowship covering statistical analysis, machine learning, software engineering, and working with data at scale. • Built neuralArt.io, an application to classify fine art paintings that could identify a painting’s artist and its genre (portrait or landscape) with 80% F1 score. Project involved scraping 2,000+ high-resolution paintings, building an image pipeline to perform image feature extractions & transformations, modeling & testing various supervised learning models such as Neural Networks, and building a web application. • Practiced applying state-of-the-art data science techniques to real-world data sets, including supervised and unsupervised machine learning, dimensionality reduction, and natural language processing. • Built a fraud prediction model for an event ticketing site with a boosted tree classifier, and developed movie recommendation system for users.

Building NeuralArt.io (Fine Art Painting Classification Application)

Built neuralArt.io, an application to classify finene art paintings that could identify a painting's artist and its genre (portrait or landscape) with 80% F1 score. Project involved scraping 2,000+ high-resolution paintings, building an image pipeline to perform image feature extractions & transformations, modeling & testing various supervised learning models such as Neural Networks, and building a web application.

September 2012 - March 2013 (7 Months)

Data Modeling Analyst

Raleigh, NC, US

• Mined Valencell’s biometric database for useful time-correlated metrics and developed bioinfor- matic models that correlate time-correlated metrics with selected health and fitness assessment. • Implemented regression and neural network models through MATLAB to design predictive stride length models and speed models for runners. • Discovered the necessity for creating seperate predicitive stride length models for trail running versus road running.

September 2013 - September 2014 (1 Year, 1 Month)

MSc/ Applied & Computational Mathematics

Oxford, England, United Kingdom

The MSc course provides extensive training in numerical linear algebra, numerical analysis of ODEs and PDEs, ODE and PDE boundary value problems and finite element methods. The case studies in mathematical modelling and scientific computing have equipped me with the tools necessary to take a well-posed problem from its initial state through its mathematical formulation and analysis to the stage where both qualitative and quantitative results can be found. For my master’s dissertation, I worked on a research project in Ice Sheet Flow Modelling. For the project, I implemented finite element methods to solve the Blatter-Pattyn equations on a continental scale and employed adjoint methods combined with inverse optimization to infer the basal friction underneath an ice sheet.

August 2008 - May 2012 (3 Years, 10 Months)

BS/ Mathematics

Durham, NC, US

My undergraduate career was heavily focused in Applied Mathematics and Mathematical Modeling. I also took extensive coursework in computer science and chemistry for my own intellectual curiousity.