Karen J Smiley | University of Pittsburgh (original) (raw)
Ms. Karen Smiley is an AI-savvy engineering leader and an advocate and independent thinker/writer on #EthicalAI, #DiversityInTech, #PeopleFirstCulture, and technology.
As a creative collaborative technology leader, she has built broad and deep global experience in the full development lifecycle for multi-platform hardware-software systems. She has built and led multiple geographically-distributed and remote agile teams that design and deliver advanced industrial analytics and toolsets.
Karen is an accomplished mentor, teacher, author, manager, coach, architect, and data scientist who is always looking for opportunities to use technology and data for good. Everything she publishes is 100% human-authored. She still enjoys creating software, analyzing data, and above all, meeting challenges and solving problems.
To date, Karen has co-invented on 12+ US patents for software technology innovations, and has authored or co-authored 40+ academic and industrial publications.
She is currently working with selected collaborators on publishing guidebooks for executives, architects, and practitioners on embedding intelligent features into software development practices, products, and people processes (to be published in 2024; look for previews on Substack).
less
Related Authors
North Carolina State University
Pontifícia Universidade Católica do Rio Grande do Sul (PUCRS)
Uploads
Papers by Karen J Smiley
Code review and inspection techniques are considered vital for defect detection during analysis a... more Code review and inspection techniques are considered vital for defect detection during analysis and design. Automated static code analyzers are essentially an approach to performing code reviews and inspections in an efficient and timely manner. Automated testing techniques are also gaining popularity and prestige. We explore various readily available code inspection tools and automated testing tools and apply a selected few to a sample application, in an attempt to evaluate the strengths and weaknesses of each tool. We then compare these results to the results obtained from execution of two suites of automated dynamic tests. Our findings indicate that these two methods have innate complementary strengths, suggesting that the joint use of both approaches (static and dynamic) can be far more effective than using either one alone.
arXiv (Cornell University), Feb 16, 2016
A single vendor cannot provide complete Industrial Internet of Things (IIoT) end-to-end solutions... more A single vendor cannot provide complete Industrial Internet of Things (IIoT) end-to-end solutions because cooperation is required from multiple parties. Therefore, interoperability is a key architectural quality. Composability of capabilities, information and configuration is the prerequisite for interoperability, supported by a data storage infrastructure and defined set of service interfaces to build applications. Secure collection, transport and storage of data and algorithms are expectations for collaborative participation in any IIoT solution. Participants require control of their data ownership and confidentiality. We propose an Internet of Things, Services and People (IoTSP) application development and management framework which includes components for data storage, algorithm design and packaging, and computation execution. Applications use clusters of platform services, organized in tiers, and local access to data to reduce complexity and enhance reliable data exchange. Since communication is less reliable across tiers, data is synchronized between storage replicas when communication is available. The platform services provide a common ecosystem to exchange data uniting data storage, applications, and components that process the data. Configuration and orchestration of the tiers are managed using shared tools and facilities. The platform promotes the data storage components to be peers of the applications where each data owner is in control of when and how much information is shared with a service provider. The service components and applications are securely integrated using local event and data exchange communication channels. This tiered architecture for composable applications reduces the cyber security attack surface and enables individual tiers to operate autonomously, while addressing key interoperability concerns. We present our framework using predictive maintenance for power transformers as an example, and evaluate compatibility of our vision with an emerging set of standards.
Code review and inspection techniques are considered vital for defect detection during analysis a... more Code review and inspection techniques are considered vital for defect detection during analysis and design. Automated static code analyzers are essentially an approach to performing code reviews and inspections in an efficient and timely manner. Automated testing techniques are also gaining popularity and prestige. We explore various readily available code inspection tools and automated testing tools and apply a selected few to a sample application, in an attempt to evaluate the strengths and weaknesses of each tool. We then compare these results to the results obtained from execution of two suites of automated dynamic tests. Our findings indicate that these two methods have innate complementary strengths, suggesting that the joint use of both approaches (static and dynamic) can be far more effective than using either one alone.
ABB Review Journal, 2019
World of simulation Simulations and digital twins are revolutionizing the way we think about the ... more World of simulation Simulations and digital twins are revolutionizing the way we think about the development and deployment of products. computer literacy, but also improvements in automation and the tools themselves, that today many such tasks can easily be performed by junior staff or students. Far from implying a de-skilling, this shift means simulations experts can now concentrate on the interpretation of results and on guiding and advising design decisions →2.
Worldwide demand for a reliable and sustainable supply of renewable energy, including solar, is g... more Worldwide demand for a reliable and sustainable supply of renewable energy, including solar, is growing. Accurate estimates of solar energy production and insights into solar equipment performance help solar plant owners and operators optimize inspections, schedule maintenance, improve the operational performance of their equipment, and maximize the environmental benefit of their investments in renewable energy. However, due to the uncertainties inherent in the unpredictable nature of this renewable resource, many challenges are associated with estimation of solar power production and detection of
INSIGHT, 2018
This paper summarizes the initial experiences and lessons learned while enabling and practicing a... more This paper summarizes the initial experiences and lessons learned while enabling and practicing agile methods at an industrial research lab. Lab staff are highly-educated personnel who innovate in complex mixed hardware/software systems, software-intensive systems, hardware components, and systems of systems. We discuss our goals in introducing agile, the challenges we faced with globally distributed industrial research in our capricious, uncertain, risky, variable, and evolving (CURVE) domains, how we addressed these challenges, results and lessons learned to date, and our next steps.
Proceedings of the International Workshop on Continuous Software Evolution and Delivery - CSED '16, 2016
Rapid delivery strategies strive to balance critical performance qualities vs. reducing the time ... more Rapid delivery strategies strive to balance critical performance qualities vs. reducing the time between an idea and deployment of a software implementation of that idea. For industrial software solutions that encapsulate expertise in deliverable components, technical SMEs (Subject Matter Experts) with ideas and knowledge have traditionally partnered as requirements providers with software development teams. These human processes are not optimally fast, are vulnerable to errors in translating or interpreting requirements, and do not scale when software teams need to integrate the knowledge of many SMEs into multiple software solutions and deployments. To address these limitations, ABB has pursued an industrial research initiative for innovative SME toolsets with focus on two goals: to accelerate the creation, evolution, reuse, and delivery of expert algorithms, and to streamline the deployment of these algorithms into releases and fielded solutions. The vision underpinning the initiative is to empower technical SMEs as “end-user developers” to convert their knowledge into reusable software solution components without having to learn, perform, or partner on traditional software development, integration, or deployment. In this paper, we summarize our experiences and lessons learned to date from this initiative, key continuing challenges, and some positional thoughts on how end-user development by technical SMEs aligns with emerging approaches for rapid delivery and evolution.
Proceedings of the 19th International Conference on Software Product Line - SPLC '15, 2015
This paper focuses on an industrial experience with software product lines of analytics-enabled s... more This paper focuses on an industrial experience with software product lines of analytics-enabled solutions, specifically the evolution of the software product line architecture for a Subject Matter Expert Workbench toolset which supports analytic plugins for multiple software product lines. As context, the toolset product line was intended for integration of expert knowledge into a family of industrial asset health applications at runtime. The toolset architecture is now being evolved to build and manage plugins for multiple Industrial Analytics solutions (software systems and services) beyond asset health. This evolution is driving changes in the desired architecture qualities of the toolset; widening the stakeholder pool and influencing priorities; affecting the architecture tradeoffs and decisions; and triggering updates to the product line architecture, the guidance for applying it, and the current prototype of the toolset. We describe our experiences in handling this evolution, assess lessons learned, and discuss potential relevance to other product line scenarios.
Lecture Notes in Business Information Processing, 2010
Global software development (GSD) is a growing phenomenon in industry, including the ABB Group of... more Global software development (GSD) is a growing phenomenon in industry, including the ABB Group of companies, which has a long history of executing globally distributed development projects. Geographic and temporal separation, culturally-based misunderstandings, and language effects are welldescribed complications for GSD teams. These factors aggravate issues (on both a practical and a leadership level) in communication, trust, and coordination, impeding the effective sharing and management of team knowledge, and creating risks to project success. In the interest of continually improving our business performance, ABB has joined the research community in exploring these issues and ways to increase awareness and tactical support for GSD project managers. In this paper, we present aggregate findings from qualitative interviews with people across different sites in the organization, and describe how identifying, measuring, and actively managing GSD-related risks can help project managers and leaders in planning and executing projects more effectively.
Jul 2, 2012 International Conference on Software Engineering and Knowledge Engineering (SEKE 2012... more Jul 2, 2012 International Conference on Software Engineering and Knowledge Engineering (SEKE 2012). Lead author Aldo Dagnino, co-author Lakshmi Ramachandran. Abstract: Fault events in a power distribution grid and substations can cause in costly power outages. Forecasting these fault events can reduce response time and enhance preparedness to repair the outage, which result in significant cost savings. Identification of fault events in distribution grids has been mostly a reactive and manual process with a relatively low level of automation. For this reason, any tools that can automate the diagnostics or prediction of fault events in the grid are welcome in the industry. The objective of the investigation presented in this paper was to develop machine-learning models capable of predicting fault events and their location in distribution grids. Historical fault event, grid electrical values, infrastructure type, and historical weather data were combined to create the predictive models. A variety of machine learning algorithms such as Neural Networks, Support Vector Machines, Recursive Partitioning, and Naïve Bayes were utilized. Neural Network models performed best at forecasting fault events given certain weather conditions, and identifying the specific grid zone where a fault occurred. The Recursive Partitioning models were better at predicting the substation and feeder where a fault occurred. An implementation at a US utility was completed to demonstrate the capabilities of these models.
Sustainable results Corporations must adapt rapidly to changing markets and adopt new technologie... more Sustainable results Corporations must adapt rapidly to changing markets and adopt new technologies to remain competitive. Such adaptations are particularly important in a rapidly changing economic climate. Flexibility and a willingness to change are important qualities that must be fostered and encouraged, at all levels, if businesses are to respond effectively to shifts in product demands or to altered customer requirements. To encourage a positive environment for such change, the process for change must be carefully planned, well managed, properly justified and applied with sensitivity. ABB uses the IDEAL SM 1.0 model as a framework to guide improvement processes so that effective changes are deployed efficiently.
At ABB, we seek to empirically evaluate the effectiveness of new collaboration tools and methods ... more At ABB, we seek to empirically evaluate the effectiveness of new collaboration tools and methods for globally distributed software development (GSD) teams. We found it challenging to identify key metrics that would serve to associate improved collaboration with improved team performance. We used the Goal-QuestionMetric (GQM) method to define a set of metrics for collaboration in GSD projects. Experiments will be conducted applying these metrics to a diverse set of development projects at ABB. Future publications will describe the outcomes of our studies at ABB.
Code review and inspection techniques are considered vital for defect detection during analysis a... more Code review and inspection techniques are considered vital for defect detection during analysis and design. Automated static code analyzers are essentially an approach to performing code reviews and inspections in an efficient and timely manner. Automated testing techniques are also gaining popularity and prestige. We explore various readily available code inspection tools and automated testing tools and apply a selected few to a sample application, in an attempt to evaluate the strengths and weaknesses of each tool. We then compare these results to the results obtained from execution of two suites of automated dynamic tests. Our findings indicate that these two methods have innate complementary strengths, suggesting that the joint use of both approaches (static and dynamic) can be far more effective than using either one alone.
arXiv (Cornell University), Feb 16, 2016
A single vendor cannot provide complete Industrial Internet of Things (IIoT) end-to-end solutions... more A single vendor cannot provide complete Industrial Internet of Things (IIoT) end-to-end solutions because cooperation is required from multiple parties. Therefore, interoperability is a key architectural quality. Composability of capabilities, information and configuration is the prerequisite for interoperability, supported by a data storage infrastructure and defined set of service interfaces to build applications. Secure collection, transport and storage of data and algorithms are expectations for collaborative participation in any IIoT solution. Participants require control of their data ownership and confidentiality. We propose an Internet of Things, Services and People (IoTSP) application development and management framework which includes components for data storage, algorithm design and packaging, and computation execution. Applications use clusters of platform services, organized in tiers, and local access to data to reduce complexity and enhance reliable data exchange. Since communication is less reliable across tiers, data is synchronized between storage replicas when communication is available. The platform services provide a common ecosystem to exchange data uniting data storage, applications, and components that process the data. Configuration and orchestration of the tiers are managed using shared tools and facilities. The platform promotes the data storage components to be peers of the applications where each data owner is in control of when and how much information is shared with a service provider. The service components and applications are securely integrated using local event and data exchange communication channels. This tiered architecture for composable applications reduces the cyber security attack surface and enables individual tiers to operate autonomously, while addressing key interoperability concerns. We present our framework using predictive maintenance for power transformers as an example, and evaluate compatibility of our vision with an emerging set of standards.
Code review and inspection techniques are considered vital for defect detection during analysis a... more Code review and inspection techniques are considered vital for defect detection during analysis and design. Automated static code analyzers are essentially an approach to performing code reviews and inspections in an efficient and timely manner. Automated testing techniques are also gaining popularity and prestige. We explore various readily available code inspection tools and automated testing tools and apply a selected few to a sample application, in an attempt to evaluate the strengths and weaknesses of each tool. We then compare these results to the results obtained from execution of two suites of automated dynamic tests. Our findings indicate that these two methods have innate complementary strengths, suggesting that the joint use of both approaches (static and dynamic) can be far more effective than using either one alone.
ABB Review Journal, 2019
World of simulation Simulations and digital twins are revolutionizing the way we think about the ... more World of simulation Simulations and digital twins are revolutionizing the way we think about the development and deployment of products. computer literacy, but also improvements in automation and the tools themselves, that today many such tasks can easily be performed by junior staff or students. Far from implying a de-skilling, this shift means simulations experts can now concentrate on the interpretation of results and on guiding and advising design decisions →2.
Worldwide demand for a reliable and sustainable supply of renewable energy, including solar, is g... more Worldwide demand for a reliable and sustainable supply of renewable energy, including solar, is growing. Accurate estimates of solar energy production and insights into solar equipment performance help solar plant owners and operators optimize inspections, schedule maintenance, improve the operational performance of their equipment, and maximize the environmental benefit of their investments in renewable energy. However, due to the uncertainties inherent in the unpredictable nature of this renewable resource, many challenges are associated with estimation of solar power production and detection of
INSIGHT, 2018
This paper summarizes the initial experiences and lessons learned while enabling and practicing a... more This paper summarizes the initial experiences and lessons learned while enabling and practicing agile methods at an industrial research lab. Lab staff are highly-educated personnel who innovate in complex mixed hardware/software systems, software-intensive systems, hardware components, and systems of systems. We discuss our goals in introducing agile, the challenges we faced with globally distributed industrial research in our capricious, uncertain, risky, variable, and evolving (CURVE) domains, how we addressed these challenges, results and lessons learned to date, and our next steps.
Proceedings of the International Workshop on Continuous Software Evolution and Delivery - CSED '16, 2016
Rapid delivery strategies strive to balance critical performance qualities vs. reducing the time ... more Rapid delivery strategies strive to balance critical performance qualities vs. reducing the time between an idea and deployment of a software implementation of that idea. For industrial software solutions that encapsulate expertise in deliverable components, technical SMEs (Subject Matter Experts) with ideas and knowledge have traditionally partnered as requirements providers with software development teams. These human processes are not optimally fast, are vulnerable to errors in translating or interpreting requirements, and do not scale when software teams need to integrate the knowledge of many SMEs into multiple software solutions and deployments. To address these limitations, ABB has pursued an industrial research initiative for innovative SME toolsets with focus on two goals: to accelerate the creation, evolution, reuse, and delivery of expert algorithms, and to streamline the deployment of these algorithms into releases and fielded solutions. The vision underpinning the initiative is to empower technical SMEs as “end-user developers” to convert their knowledge into reusable software solution components without having to learn, perform, or partner on traditional software development, integration, or deployment. In this paper, we summarize our experiences and lessons learned to date from this initiative, key continuing challenges, and some positional thoughts on how end-user development by technical SMEs aligns with emerging approaches for rapid delivery and evolution.
Proceedings of the 19th International Conference on Software Product Line - SPLC '15, 2015
This paper focuses on an industrial experience with software product lines of analytics-enabled s... more This paper focuses on an industrial experience with software product lines of analytics-enabled solutions, specifically the evolution of the software product line architecture for a Subject Matter Expert Workbench toolset which supports analytic plugins for multiple software product lines. As context, the toolset product line was intended for integration of expert knowledge into a family of industrial asset health applications at runtime. The toolset architecture is now being evolved to build and manage plugins for multiple Industrial Analytics solutions (software systems and services) beyond asset health. This evolution is driving changes in the desired architecture qualities of the toolset; widening the stakeholder pool and influencing priorities; affecting the architecture tradeoffs and decisions; and triggering updates to the product line architecture, the guidance for applying it, and the current prototype of the toolset. We describe our experiences in handling this evolution, assess lessons learned, and discuss potential relevance to other product line scenarios.
Lecture Notes in Business Information Processing, 2010
Global software development (GSD) is a growing phenomenon in industry, including the ABB Group of... more Global software development (GSD) is a growing phenomenon in industry, including the ABB Group of companies, which has a long history of executing globally distributed development projects. Geographic and temporal separation, culturally-based misunderstandings, and language effects are welldescribed complications for GSD teams. These factors aggravate issues (on both a practical and a leadership level) in communication, trust, and coordination, impeding the effective sharing and management of team knowledge, and creating risks to project success. In the interest of continually improving our business performance, ABB has joined the research community in exploring these issues and ways to increase awareness and tactical support for GSD project managers. In this paper, we present aggregate findings from qualitative interviews with people across different sites in the organization, and describe how identifying, measuring, and actively managing GSD-related risks can help project managers and leaders in planning and executing projects more effectively.
Jul 2, 2012 International Conference on Software Engineering and Knowledge Engineering (SEKE 2012... more Jul 2, 2012 International Conference on Software Engineering and Knowledge Engineering (SEKE 2012). Lead author Aldo Dagnino, co-author Lakshmi Ramachandran. Abstract: Fault events in a power distribution grid and substations can cause in costly power outages. Forecasting these fault events can reduce response time and enhance preparedness to repair the outage, which result in significant cost savings. Identification of fault events in distribution grids has been mostly a reactive and manual process with a relatively low level of automation. For this reason, any tools that can automate the diagnostics or prediction of fault events in the grid are welcome in the industry. The objective of the investigation presented in this paper was to develop machine-learning models capable of predicting fault events and their location in distribution grids. Historical fault event, grid electrical values, infrastructure type, and historical weather data were combined to create the predictive models. A variety of machine learning algorithms such as Neural Networks, Support Vector Machines, Recursive Partitioning, and Naïve Bayes were utilized. Neural Network models performed best at forecasting fault events given certain weather conditions, and identifying the specific grid zone where a fault occurred. The Recursive Partitioning models were better at predicting the substation and feeder where a fault occurred. An implementation at a US utility was completed to demonstrate the capabilities of these models.
Sustainable results Corporations must adapt rapidly to changing markets and adopt new technologie... more Sustainable results Corporations must adapt rapidly to changing markets and adopt new technologies to remain competitive. Such adaptations are particularly important in a rapidly changing economic climate. Flexibility and a willingness to change are important qualities that must be fostered and encouraged, at all levels, if businesses are to respond effectively to shifts in product demands or to altered customer requirements. To encourage a positive environment for such change, the process for change must be carefully planned, well managed, properly justified and applied with sensitivity. ABB uses the IDEAL SM 1.0 model as a framework to guide improvement processes so that effective changes are deployed efficiently.
At ABB, we seek to empirically evaluate the effectiveness of new collaboration tools and methods ... more At ABB, we seek to empirically evaluate the effectiveness of new collaboration tools and methods for globally distributed software development (GSD) teams. We found it challenging to identify key metrics that would serve to associate improved collaboration with improved team performance. We used the Goal-QuestionMetric (GQM) method to define a set of metrics for collaboration in GSD projects. Experiments will be conducted applying these metrics to a diverse set of development projects at ABB. Future publications will describe the outcomes of our studies at ABB.
INFORMS Data Science Workshop, 2019
Worldwide demand for a reliable and sustainable supply of renewable energy, including solar, is g... more Worldwide demand for a reliable and sustainable supply of renewable energy, including solar, is growing. Accurate estimates of solar energy production and insights into solar equipment performance help solar plant owners and operators optimize inspections, schedule maintenance, improve the operational performance of their equipment, and maximize the environmental benefit of their investments in renewable energy. However, due to the uncertainties inherent in the unpredictable nature of this renewable resource, many challenges are associated with estimation of solar power production and detection of performance issues.
In this study, our goal is to explore how predictions of solar inverter and plant production can be
improved by applying data science techniques, and how machine learning models can be applied to
correctly classify malfunction causes for solar inverters. Our results show that regional weather data can be used to estimate (and potentially predict) solar energy production for some applications; that a hybrid machine learning model based on historical data, temperature, and information from physical models outperforms predictions from state-of-the-art physical models; and that environmental factors such as lightning and ambient temperature, as well as grid operating conditions, can influence device reliability.
Authors: Mauricio Soto, Karen Smiley, Xiao Qu, Travis Galoppo, Alok Kucheria, Rohini Kapoor, Melwin Jose
Proceedings of the Software Engineering Process Group (SEPG) North America 2010 Conference, 2010
It is well known that fixing requirements errors which escape into later phases of the software d... more It is well known that fixing requirements errors which escape into later phases of the software development phases can be expensive and time-consuming. Measuring requirement processes and quality early can help project managers and requirements analysts better understand requirements-related risks and address the potential impact to the project. This presentation describes our experiences and challenges in defining and applying Requirements Engineering (RE) measurements in four categories: quantity metrics, quality metrics, volatility metrics, and process metrics.