Nataliya Portman

• Developed content recommendation system for financial institution based on probability concepts that delivers right message to right audience
• Developed and implemented methodology for evaluation of business impact of adopting recommendation model by financial client
• Validated recommendation system intelligence and won 6-figure business deal with future opportunities for software upgrade and revenue growth
• Created Microsoft PowerBI visualization tool for audience discovery that assists financial clients in planning and timely management of Digital-Out-Of-Home advertising campaigns
• Supported internal sales teams by bringing in relevant web-scraped data on currently active mall tenants and their categories to PowerBI dashboard providing guidance in ad placement across malls for retail clients
• Established thought leadership in Data Science approaches for recommendation systems in Digital-Out-Of-Home industry through participation in professional machine learning conferences
• Implemented Markov Chain Monte-Carlo methods for stochastic modelling of traffic counts in front of Digital-Out-Of-Home ads (in terms of numbers of people passing by screens) and traffic forecasting needed for pricing of advertising services in malls and movie theatres.
• Extracted and assessed data from CRM databases to drive improvement of product development and business strategies and processes

• Created materials and taught course on Advanced Mathematics of Machine Learning (generative models for topic discovery and image processing methods),
• Created materials and taught course on Mathematics of Artificial and Recurrent Neural Networks.

• Spearheaded sales channel forecasting projects that helped channel leaders maximize ROI of their channel incentive programs. Specifically, developed and implemented uncertainty-aware forecasting methodology (based on SARIMAX statistical model) that
  (1) assists in budgeting of rebate programs
  (2) evaluates marketing interventions and key product sales
  (3) verifies alignment with go-to-market strategy
  (4) forecasts numbers of reward winners based on predicted annual performance
  
• Analyzed loyalty program member data and delivered insights on customer retention
  
• Developed statistical framework for evaluation of effectiveness of training programs for sales associates
• Established good rapport with stakeholders via regular reviews and updates of forecasting models
• Introduced continuous improvement approach to predictive analytics services with focus on accuracy
• Supported Business Intelligence developers via assembling data from various sources (SQL database, Excel files, Data Warehouse), data cleaning and shaping data into schema suitable for calculation of KPIs and visualization of data segments in Qlik Sense
• Hired highly qualified personnel and talented intern students to increase efficiency of Insights team work
• Orchestrated team building and professional achievement celebration events to strengthen team spirit and to show value of team member contributions
• Established Insights request intake process and dynamic Insights Product Roadmap that captures task priority change and informs team members on their performance

• Pioneered innovative Data Science projects aligned with organizational business objectives that resulted in increase of company's share prices
• Led small team (two Data Engineers and junior Data Scientist) in early stages of development
• Introduced vehicle videodata acquisition protocol and designed end-to-end data processing pipeline on GCP
• Built Tensorflow deep learning models based on collected video samples for recognition of target vehicle views and prepared them in format ready for deployment on mobile device
• Established model validation process that involves testing video data acquisition, training data preparation and retraining of Inception-V3 classifier on updated (bigger) video dataset followed by calculation of accuracy metrics
  
• Worked closely with data engineers to launch used car pricing models via model selection approach and system for monitoring predictive model accuracy
  
• Created content-based recommender system based on Random Forests algorithmic models of buyer’s preferences
  
• Co-invented and patented three innovative Data Science products ("BidAssist" (strategy for setting up initial bidding amount), Mobile AutoVision, Used Vehicle Price Estimator)

• Developed computationally efficient algorithms for validation of supervised learning models of ferromagnetic system behavior (using Markov Chain Monte-Carlo methods)
  
• Contributed to Neural Network modelling of agent-based communication systems and performed clustering analysis of simulated human learning behavior data that led to better understanding of dynamic agent responses
• Contributed to ACM’s Women in Computing organization by running fundraising and Data Science educational events for UOIT students

• Advised in research and development of new methods to analyze and diagnose dementia using electroencephalogram (EEG) patient data and company-developed software for dynamic electrical cortical imaging.
  

• Developed and implemented novel machine learning algorithm for recognition of common ingredients causing food sensitivities and nutrient estimation from near-infrared spectra of foods
  
• Developed validation tools in Python for performance evaluation of regression and classification models
  
• Wrote white paper for marketing TellSpec technology and received high praise from management for impressive performance of food content recognition algorithm at Consumer Electronics Show in Las Vegas
  

Contributed to the MRI study of early normal brain development funded by the National Institutes of Health.

Projects:

• Image fusion algorithm and its role in longitudinal MRI study of early brain development
  

— Developed a multi-resolution image fusion algorithm that effectively combines information from different MR brain image types into a single image with enhanced anatomical features

— Created representative fused templates of pediatric MR dataset of early brain development and showed a role of image fusion in longitudinal study.

• Local semi-supervised and perceptual image quality-based approach to automatic brain tissue classification in young children
  

— Introduced a new philosophy of brain tissue classification without ground truth using modern pattern recognition and perceptual image quality models in Computer Vision to allow detection of tissue patterns with an accuracy of the Human Visual System

— Developed a regional version of the proposed classification approach that accommodates highly varying within-class MR signal intensities and improves detection of small brain structures in low-contrast regions

— Designed and analyzed an algorithm based on a novel regional classification methodology and implemented it to pediatric MR dataset of early brain development

— Extended the methodology to identification of myelinated and unmyelinated white matter subclasses in early infancy period from 0 to 6 months of age.

• Published PhD thesis-related paper and book under title “The Modelling of Biological Growth: A Pattern Theoretic Approach”
  

• Planned and delivered lectures, developed several exam problems and graded exams for

1. Calculus 2 for Honours Mathematics Spring 2007

2. Calculus 1 for Engineering Winter 2009

Guided students in solving assignment problems and graded assignments and exams for

• Calculus 3 for Honors Mathematics Winter 2006
• Calculus 2 for Engineering Winter 2008
• Computational Methods for Differential Equations Spring 200
• Applied Complex Analysis Spring 2008

Conducted weekly meetings with young female professionals focused on

• identification of professional goals and how to achieve them
• successful transitioning to a leadership role
• strategies to address challenging situations in the workplace