Primary Professional Experience

2018-current, Senior Computer Scientist

NEURIO TECHNOLOGIES INC. – Vancouver, BC, Canada

Sample Projects

• Forecasting home electricity consumption and generation – Applying machine learning algorithms using time series of sensor and weather data to predict an individual home's electricity profile in the coming hours/days, which is then used to optimize the operation of a storage system (e.g. battery) – Helping design and implement the software architecture to support data retrieval and processing, model training, prediction and evaluation, and whole system performance monitoring.

2006-2017, Senior Complexity Scientist

ICOSYSTEM CORP. – Cambridge, MA, USA

General Responsibilities

• Work independently or as part of a team on client and internal R&D projects
• Act as technical lead for small teams
• Participate in client meetings, learn about the client's domain, build common ground, and help identify the client's needs
• Scope out project requirements, time estimates for delivery and resource needs
• Identify relevant computational, mathematical and statistical techniques
• Identify potential data sources, from the client or public
• Develop end-to-end solutions: build models, incorporate data, design and implement algorithms, run simulations, perform data analyses, produce visualizations
• Interact with the client on an ongoing basis, building their trust and understanding of the solution
• Present results of complex analyses in simple actionable terms to non-technical business audiences
• Produce technical documentation
• Provide technical support to the business development team
• Conduct research and development for the generation of intellectual property suitable for patenting, licensing or sales
• Keep up to date with the state-of-the-art in relevant scientific fields

Sample Projects

• Design of Resilient Water-Distribution Networks – Developed and applied evolutionary and co-evolutionary algorithms to automate the placement of smart-valves in ship-based piping networks, so as to optimize damage/fault resilience; also to automate the design of control algorithms – Conducted theoretical research into co-optimization algorithms  – Published results in peer-reviewed scientific conferences and journals – To be transitioned from applied research to technology development by the US Office of Naval Research.
• Healthcare Data Analysis and Visualization – Conducted exploratory data analysis of US Medicare-released data to gain insights into hospital usage. Developed interactive visualizations for navigating the data – Conducted statistical analysis of proprietary patient data to identify potential drivers of cancer survivability (Client: large international pharmaceuticals company).
• Personnel Planning - Developed fast constraint optimization heuristics for grouping work tasks on Navy ships in a way that minimizes costs while improving schedule desirability for sailors; compared to previous algorithms, reduced runtime by as much as two orders of magnitude while also improving solution quality.
• Blog-Network Analysis – Performed connectivity analysis and text mining for a large internet network of food-related blogs; implemented network clustering algorithms to detect communities of blogs and visualized results – Used by large food manufacturing company to assess online opinion of their brand, and which blogs are most influential.
• Fisheries Supply Chain Modeling – Co-developed open-source agent-based model and web-based simulation of the operation of a value chain moving tuna from a small-scale fishery all the way to consumers, through intermediaries and markets – Analyzed the impact of hypothetical interventions, to assist the Rockefeller Foundation with decision-making for their initiatives in support of sustainability of fish stocks and maintaining livelihoods of traditional tuna fishers.
• Text Visualization – Co-developed algorithms for multi-source text analysis and visualization, including extraction of salient features in the form of a network of words, detection of network clusters, and computation of network layout for display – Co-author on US patent – Technology was spun off as a start-up company (Infomous.com), whose clients include The Economist and Washington Post, and whose dynamic word “clouds” have been featured on blogs, news and social network sites all over the world.
• Strategic Portfolio Planning – Co-developed agent-based simulation of the operation of a drug-development pipeline – Developed evolutionary algorithms to automate the search for optimal portfolio strategies (fiscal, resource acquisition and allocation, R&D, portfolio management, deal making) – Designed and recorded screencast tutorials for the software – Resulting tool is actively used by large pharmaceutical company for long-range planning.
• Route Optimization – Developed dynamic programming algorithm for optimizing traversal order of street segments on a postal route – Incorporated in tool actively used for route design by the French postal system.
• Fuel Pricing – Performed statistical analysis to investigate drivers of competitor pricing strategies – Developed evolutionary algorithms to predict unusual price movement – Client: large international petroleum company.
• Financing Term-sheet Optimization – Developed evolutionary algorithms to search for term-sheets optimizing fairness of economic outcomes (e.g. capitalization tables) under multiple scenarios (e.g. IPO, liquidation) – Incorporated in tool built in collaboration with The Law Lab at Harvard University.
• Contextual Advertising – Developed software architecture and information retrieval algorithms to match relevant ads from a database to words in a network-based text visualization (see above).
• Consumer Decision-Making – Co-developed simulation of how US seniors choose a private health-insurance plan, to help insurance companies drive plan design by predicted market share – Implemented evolutionary algorithms to calibrate model parameters using historic data – Resulting tool sold to large US health-insurance company (Humana).
• Market Access – Co-developed simulation of the process whereby typical decision committees at private health-insurance companies select drugs to cover; developed evolutionary algorithms to calibrate model parameters using historic data – Resulting tool is actively used by large drug-development company to predict the likelihood that a given drug in early clinical testing stages will be covered, once on the market, by various health-insurance companies.