As a Staff Data Scientist, I exercise my advanced data science expertise designing, developing, and deploying models and analysis for approximately 70% of the time on specific high impact projects, and perform technical project management and consulting for of data-driven initiatives about 30% of the time. Sometimes this split is 50%-50% if the group has lots of new modeling and analysis tasks for a new initiative that need to be technically planned out and verified. Job Responsibilities:- Advanced Implementation Responsibilities: * Actively researching, developing, implementing, and inventing the most advanced algorithms/models/analysis on a select few projects * Participating in design reviews and recommend improvements * Assisting in risk assessment and mitigation activities * Working with other data scientists in planning, prioritizing and executing assigned tasks within deadlines- Leadership Responsibilities: * Leading a team of 8 data scientists on technical aspects of data science initiatives * Conducting strategic technical data science planning * Recommending new technologies to ensure quality and productivity- Communication Responsibilities: * Sharing expertise and insight with other data scientists * Motivating data science staff and fostering collaboration * Writing technical papers on data science work being done * Writing technical and non-technical blogs on work being done

My time is spent researching, developing, and deploying (That's right, all 3. I deploy my research so it can have a real, tangible business impact. Groups that perform R&D and have consistent and automated deployment frameworks have a huge competitive advantage) ML and AI (And I mean real agent based, decision making systems) for many major areas of Cisco's customer experience including:Customer Journey Predictions: *Using unsupervised (Clustering, outlier detection, etc...), supervised (Classification and Regression models using every kind of ML algorithm), and semi-supervised methods automatically (along with AutoML) for producing deep insights about how customers interact with cisco (such as likelihood for a customer to move from one category of behavior interaction to another)Prescriptive Actions: *Using statistical analysis methods from causal inference literature, agent/utility theory, bayesian decision analysis, reinforcement learning, and continuous optimization/testing methods to create a pipleine that prescribes business actions of the form "If you take action X, we predict and improvement in metric Y by Z%".ConTEXT AI: *Using advanced ML algorithms, along with cutting edge deep learning methods for categorizing and analyzing sets of customer interactions online, analyzing and retrieving related text resources, and selecting/generating responses using deep learning based NLG/Information Extraction methods.Sales Opportunity Analysis: * Using ML models and BigData frameworks (such as spark and hive) to process terabytes of sales opportunities to identify promising opportunities.I also head a deep NLG/NLP research reading group, am a technical content manager for the 'AI for CX Blog' which can be found here "aiforcxblog.com", and give internal talks at Cisco about everything from how to use GCP and cloud services for AI projects and workloads, to the latest methods for machine learning and deep learning for NLP/NLG and beyond.

My time is spent researching, developing, and deploying (That's right, all 3. I deploy my research so it can have a real, tangible business impact. Groups that perform R&D and have consistent and automated deployment frameworks have a huge competitive advantage) ML and AI (And I mean real agent based, decision making systems) for many major areas of Cisco's customer experience including:Customer Journey Predictions: *Using unsupervised (Clustering, outlier detection, etc...), supervised (Classification and Regression models using every kind of ML algorithm), and semi-supervised methods for producing deep insights about how customers interact with cisco (such as likelihood for a customer to move from one category of behavior interaction to another) *Combining the above mentioned algorithms into an AutoML pipeline for automatically generating baseline models, and for scaling to multiple product offers.Prescriptive Actions: *Using statistical analysis methods from causal inference literature, agent/utility theory, bayesian decision analysis, and continuous optimization/testing methods to create a pipleine that prescribes business actions of the form "If you take action X, we predict and improvement in metric Y by Z%".SocialAI: *Using standard ML algorithms, along with cutting edge deep learning methods for categorizing and analyzing sets of customer interactions online, and using these categorizations to select/generate responses using deep learning based NLG/Information Extraction methods.I also head a deep NLG/NLP research reading group, am a technical content manager for the 'AI for CX Blog' which can be found here "aiforcxblog.com", and give internal talks at Cisco about everything from how to use GCP and cloud services for AI projects and workloads, to the latest methods for machine learning and deep learning for NLP/NLG and beyond.

I was previously researching machine learning for irregular time series data sets, with particular validation against electronic health record data. Irregular time series can also be thought of "event based" systems, where events are occurring over time at different intervals. Examples of problems that fit this domain include log based event analysis, social media events (think tweets), operating system events, health record events, market trading events, and many others.Overall I'm interested in machine learning, but have picked up a fancy for trying to develop new methods for reasoning about data that is generated over time. I am currently working with Dr David Page's research group located in the Wisconsin Institute of Discovery.

I was previously teaching one of the sections of CS 540, also known as "Introduction to AI" at the University of Wisconsin, to roughly 200 students. This includes teaching traditional AI topics such as uninformed/informed search, continuous search, game theory, propositional and first order logic, as well as various topics from machine learning.

My responsibilities included applied machine learning, mathematical multivariate functional analysis, and scalable system design for high throughput and high computation intensive applications. Main Tools Used For Distributed, Scalable ML Software Platform:*Docker*Netflix Conductor

My research and software engineering work includes the development of:Unsupervised machine learning systems for the analysis of complex health data including:***Multiple kinds of clustering (K-means, hierarchical, density based, etc...)***Multiple kinds of nonlinear feature engineering (Kernel Learning, etc...)***Multiple kinds of nonlinear dimensionality reduction (PCA, Kernel PCA, LDA, Self Organizing Feature Maps, Nonlinear manifold embedding, etc...)Supervised machine learning systems for the analysis of complex health data including:***Statistical prediction of anomalous health events.***Time series based forecasting of health status.***Classification and categorization of individual health status.***Planning systems for health recovery based on past activity data.Full stack models and application software:***Android mobile phone applications for gathering sensor data, displaying data, and communicating with software services.***Software services for providing access to predictive models, and for application data storage. I developed many statistical prediction models for the analysis of health events, statistical analysis of latent data patterns, and designed pipelined machine learning systems that scale for large numbers of simultaneous users. My work included both the practical application of machine learning that dealt with model selection, feature engineering, model evaluation, and algorithmic efficiency, as well as the development of novel theoretical machine learning methods for increasing performance considering specialized domain centric constraints.I also worked on various software, hardware, and firmware projects. Most of these dealt with aspects of health monitoring, and new health tracking technology. This includes developing new software algorithms for quantifying health, development of wearable embedded systems, and development of Android applications for interfacing with health devices.

Research Project Results:***Custom Deep CNN and optimization method, Keras/Theano, Bloodless Blood-Glucose prediction model with 98% prediction accuracy***Feature importance analysis using random forests, LASSO, and other feature selection based methods for better insight into the problem domain.***Unsupervised analysis such as meal information clustering, and time series similarity analysis***Scalable, AWS based deep net model that could be trained with AWS based graphics card clusters***Android mobile phone and smart watch application for gathering data, communicating with Deep CNN model, and displaying user glucose level status.***Scalable AWS based micro services for managing user data, and predictive computations.My research work for the Trustworthy Data Engineering Lab centered around mining meaningful health metrics from wearable health platforms, and developing models to relate them to clinical health metrics. During my time with the lab, I developed a glucose peak prediction system using advanced machine learning methods, which makes predictions using only common wearable sensors. The system is an end to end solution, including services running in the cloud for data manipulation and ml model access, a mobile front end for displaying and gathering data, and a smart watch for displaying prediction updates.My work in general included the use of various clustering, statistical modeling, feature learning, data mining techniques, and the development of scalable machine learning algorithms. Most of the code was implemented using python, and includes the use of Scipy, Pandas, Scikit-Learn, Theano, and Spark.

As a student at UC, I studied the field of computer engineering. This includes not only taking classes and participating in clubs, but actively researching new fields of technology. Much of my research focused on sweat sensing devices, mobile health software applications, and mobile health analysis algorithms.Note that UC's engineering program is 5 years instead of 4 due to the required 5 semesters of COOPs. The extra year to graduate in exchange for so much industry experience is well worth it, and in my opinion puts UC's engineering program light-years ahead of most academic institutions which rely on student internships for exposure to industry.

Software engineering and development as a part of Intuit's small business group. I work on the payments platform developing services for managing payments. This work includes the use of numerous software development tools and frameworks. The major frameworks used include Mule ESB, Tomcat, Spring, JAXB, Jackson, and Jersey (As well as some JS specific frameworks). Other numerous frameworks are used for design, development, testing, and deployment.

My research and software engineering work includes the development of:Unsupervised machine learning systems for the analysis of complex health data including:***Multiple kinds of clustering (K-means, hierarchical, density based, etc...)***Multiple kinds of nonlinear feature engineering (Kernel Learning, etc...)***Multiple kinds of nonlinear dimensionality reduction (PCA, Kernel PCA, LDA, Self Organizing Feature Maps, Nonlinear manifold embedding, etc...)Supervised machine learning systems for the analysis of complex health data including:***Statistical prediction of anomalous health events.***Time series based forecasting of health status.***Classification and categorization of individual health status.***Planning systems for health recovery based on past activity data.Full stack models and application software:***Android mobile phone applications for gathering sensor data, displaying data, and communicating with software services.***Software services for providing access to predictive models, and for application data storage. I developed many statistical prediction models for the analysis of health events, statistical analysis of latent data patterns, and designed pipelined machine learning systems that scale for large numbers of simultaneous users. My work included both the practical application of machine learning that dealt with model selection, feature engineering, model evaluation, and algorithmic efficiency, as well as the development of novel theoretical machine learning methods for increasing performance considering specialized domain centric constraints.I also worked on various software, hardware, and firmware projects. Most of these dealt with aspects of health monitoring, and new health tracking technology. This includes developing new software algorithms for quantifying health, development of wearable embedded systems, and development of Android applications for interfacing with health devices.

This organization was aimed at introducing up an comming high school students with thechnical careers in cincinnati. As a student in this organization, I attended many seminars on technical skills. These included seminars on Android application development, web page development, and php.

As a food service manager, I:Managed food preparation tasks.Managed cash distribution to registers.Managed and trained employees for individual food carts.Devised employee task distribution.Promoted teamwork based problem solving when working in the kitchen.