I was the primary caretaker for my grandfather whose dementia has been getting worse and has been developing other significant health conditions. His situation has now stabilized and I am no longer his primary caretaker.

• Explore various data sets to answer specific questions from business/marketing, hardware, firmware, and web/app groups, then report on and discuss the implications of the results of the analyses with them• Summarize progress from the data science team for regular meetings to report to the wider department with diverse focuses on various aspects of smart thermostat design and support• Design next generation of custom energy saving algorithms for smart thermostats, focusing on balancing comfort preferences and energy saving behaviors in an automated way• Use various machine learning models (linear regression, logistic regression, boosted decision tree, neural network) to predict which replacement parts an HVAC (heating, ventilation, and air conditioning) repair tech is likely to need for a given set of system errors• Develop custom algorithms and predictive models that use data from smart thermostats to detect when components of an HVAC system need servicing, have failed, to predict when they may be near failing, or are exhibiting various other problems - Includes initial investigations to understand the given issue, prototyping an algorithm(s) to validate effectiveness for the majority of relevant cases, running over a wide range of test data to find corner cases and make adjustments so they are handled reasonably, preparing the workflow for production and determining any monitoring required

• Use data-driven approaches to bring insight to various business discussions - Conduct initial feasibility studies for new business directions to provide data-driven insight for business decisions - Quickly generate back-of-the-envelope estimates for predicting how products will perform in new conditions to support discussions with prospective partners - Review and explain analyses from both competitors and independent studies to aid discussions with prospective partners, often focusing on differences in the methodology and shortcomings in interpretation which affect the apparent results - Conduct transparent in-house M&V (measure and verify) analysis. As a cross check of poorly run third party results. Significant improvements seen in the preliminary results provide a stronger position to negotiate more advantageous payment levels. Also as energy load reduction and shifting M&V analysis for a pilot program.• Enhance smart thermostat learning algorithms to significantly improve results - Identify and correct bugs in algorithms, decreasing runtime and stabilizing results - Design and build more robust energy savings algorithms to better adapt to individual lifestyles and balance between comfort and savings - Develop a more robust algorithm to detect rapid cycling of HVAC (heating, ventilation, and air conditioning) systems - Automate fault detection for HVAC systems which eliminated nearly all labor costs from previously manual steps - Expand range of detected HVAC faults by analyzing patterns in manually identified cases• Design and build software to simulate HVAC runtime for various conditions in order to study the effects of different setpoint offsets and lengths for demand response events• Create, maintain, and manage automated processes that run regular data processing and aggregation jobs