After having spent a career using my data science, machine learning and A.I. skills to build products across a multitude of disciplines I decided to use them in the trading space. I took some time off to learn and begin trading with my own funds. After a year and a half of continuous successful trading and beating all hedge fund stats I decided to open my own hedge fund.

Providing technical advice in the fields of data science, machine learning and statistics. Both from a technical aspect and a project to product approach.

Advisor on product.

I lead the Data Science teams at IDEAs within Microsoft. The IDEAs group is positioned in a way to leverage data coming from across the main products across Microsoft. We use this data across our different groups of Data Scientists, Machine Learning Engineers and Machine Learning Scientists to leverage AI and Machine Learning to grow Microsoft products such as: XBOX, Windows, M365 and more.

When looking to bring the value of AI to an organization, it is important to consider growing the number of AI-enabled employees. One efficient way to achieve this is via software. This is the software we have built at Ople.ai. We started off with an AI platform covering everything from feature engineering, to model building and evaluation, to model deployment and usage (an AutoML platform). Once we had that covered, we decided to work on the other valuable aspects of building a model beyond making predictions. The first step was to build intuitive and accessible model explainability, this allows business users to make more sense of a model’s predictions and have more information in order to make a business decision. The next step towards bringing additional value with our platform was what we have called “Data Explainability”. This aspect of the platform uses the model to uncover valuable information about the data used to build it. We have found that data analysts and business users find tremendous value in this feature. At Ople.ai, I have gained tremendously valuable experience in turning a vision into reality. I have worked on pitching to investors and raising funds, speaking to customers to better understand their needs as well as in a sales capacity, hiring, managing and helping others grow. I also helped foster a culture where people felt passionate about the vision of the company and were motivated, encouraged and enabled to meaningfully contribute to the company’s outcome. Through a group of incredibly talented people from all backgrounds, I have also been able to learn and grow as a professional and a person.

Leading the data science efforts across the various groups and projects in the organization. Helping grow and develop the data science teams. Understanding and improving on current models that are used by customers and creating new ones to power new products.

Heading the data science and machine learning efforts at Banjo (Ban.jo), where I have grown a data science team to tackle incredibly challenging problems in the social network arena. I am building models to detect events as they happen across the world in real time as well as predicting their topics. My team and I are using deep learning (among other techniques) to extract content from images, video, and text to understand what is happening in the world everywhere and at every minute. Projects that range from detecting brands and objects in images to understanding scenes and actions. One of the projects focuses on assigning exact geo-location to social images that do not have geo information based solely on the actual content of the image.

Designing overall data mining approaches to improve outcomes through model definitions, target specifications and evaluation metrics. Providing data mining expertise in various areas to increase efficiency and effectiveness of modeling efforts, such as the examples below. Feature extraction from a rich set of data including images, free-form text, personal profiles, and online behavior. Using techniques such as: restricted boltzmann machines, deep belief networks, clustering, and dimensionality reduction. Developing, closely with engineers, parallel implementations of particle swarm optimization algorithm that will allow for fast feature selection and the creation of predictive models that can optimize non-differentiable loss functions. This implementation will allow the mimicking of various machine learning algorithms such as: logistic regression, neural networks, RBF networks, as well as ensembling methods like stacking, grading, voting, boosting, bagging, gated ensembles, greedy ensembles and cascading.

This position involved performing Data science on a variety of healthcare related problems. I was part of a small team that was responsible for creating products that would boost the overall value of the company to their clients. Work was done in an independent fashion: from project design, discovery phase, specification and requirements, research, data acquisition (databases), feature engineering and selection, training of model, final documents (report, specifications, code, sql, description, model) and knowledge transfer to engineers. I have worked on projects for predicting: hospital readmissions, health insurance future costs on individual level, health metrics, imputation of missing data, visualization of high-dimensional space, injury detection through tracking jumping mechanics, sports (soccer) strategy effectiveness evaluation, transfer to pediatric intensive care unit and others.