Mykola Hayvanovych

This project had a clear objective of developing a set of comprehensive analytical capabilities on the foundation of Hadoop’s distributed storage and processing platform. Utilizing my theoretical expertise and real-world application of Machine Learning, Data Mining, and Graph Theory, I made key and instrumental contributions in the conceptualizing, planning, and creation of the Advanced Analytic Toolkit project. Building on the analytical toolkit a customer can quickly develop customized… Show more

This project had a clear objective of developing a set of comprehensive analytical capabilities on the foundation of Hadoop’s distributed storage and processing platform. Utilizing my theoretical expertise and real-world application of Machine Learning, Data Mining, and Graph Theory, I made key and instrumental contributions in the conceptualizing, planning, and creation of the Advanced Analytic Toolkit project. Building on the analytical toolkit a customer can quickly develop customized analytic applications and shorten the path from deployment to production. The success of this project served as a key differentiation for Intel and helped to strengthen Intel’s position as a leader in Big Data Analytic space.
Key Contributions:
• Architected initial toolkit design, concepts and direction
• Singlehandedly created initial Proof of Concept
• Worked with management in order to secure funding for this project
• Developed a number of Machine Learning, Data Mining, and Graph analytics for the toolkit, utilizing Java MR-framework as well as Apache Spark with Scala. Show less

Key concept behind this project was to develop a flexible platform which would enable sponsors to easily manage heterogeneous analytics.
Key contributions:
• Performed R&D to advance the concepts of the analytical platform.
• Participated in the development of a working prototype for one of the sponsors.
• Developed and implemented one of the key parts of the platform. Its core was built using Multithreaded Algorithms and JASON RPC communication protocol.
• Developed and… Show more

Key concept behind this project was to develop a flexible platform which would enable sponsors to easily manage heterogeneous analytics.
Key contributions:
• Performed R&D to advance the concepts of the analytical platform.
• Participated in the development of a working prototype for one of the sponsors.
• Developed and implemented one of the key parts of the platform. Its core was built using Multithreaded Algorithms and JASON RPC communication protocol.
• Developed and Implemented Analytical Engine, which interacted with other services through WSDLs and exposed analytics to the rest of the platform.
• This project received invention of the year nominee award and patent pending.
Show less

The project focused on introducing machine learning sensors to improve intrusion detection capabilities of the sponsor. I was technical and team lead for this project.
Key Contributions:
• Selected resources and formed two teams consisting of nine people.
• Facilitated regular weekly interactions with the sponsor, as well as organized
Internal team discussions.
• Worked closely with the team, throughout the process of sensor architecture development.
• Selected… Show more

The project focused on introducing machine learning sensors to improve intrusion detection capabilities of the sponsor. I was technical and team lead for this project.
Key Contributions:
• Selected resources and formed two teams consisting of nine people.
• Facilitated regular weekly interactions with the sponsor, as well as organized
Internal team discussions.
• Worked closely with the team, throughout the process of sensor architecture development.
• Selected appropriate existing machine learning approaches from WEKA and MALLET, as well as proposed and implemented novel algorithms for the purposes of the project. Apache Commons Math library was utilized during implementation.
• Developed XML configurable architecture with a Multithreaded Algorithm Core.
• Developed memory efficient and extremely agile modules to extract new useful additional network features as well as to select existing ones.
• Received a special personal achievement award for initiating and executing this project.
Show less

Led a large team through implementation of the system that identifies relevant features and patterns behind diseases, their similarities and differences in order to more accurately identify suspect conditions and link ICD9/ICD10 codes with HCCs.
Key Contributions:
• Led identification and cleansing of data sources and their integration into HDFS cluster. Hive and Hbase/Accumulo were utilized for these purposes.
• Directed integration and attribution of highly multidimensional… Show more

Led a large team through implementation of the system that identifies relevant features and patterns behind diseases, their similarities and differences in order to more accurately identify suspect conditions and link ICD9/ICD10 codes with HCCs.
Key Contributions:
• Led identification and cleansing of data sources and their integration into HDFS cluster. Hive and Hbase/Accumulo were utilized for these purposes.
• Directed integration and attribution of highly multidimensional heterogeneous medical data from a number of sources
• Conducted exploratory feature analysis, feature engineering, enrichment and pruning.
• Architected creation of Graph based system, which allowed for a number of analyses: Graph Queries, commonality analysis, anomaly detection, what-if analysis and others. Show less

Led the team and was the key technical contributor in development of Machine Learning system enabled by Hadoop to extract patterns and to help identify potential new voters, donors and to optimize funds and awareness raising campaigns.
Key Contributions:
• Supervised identification of proper data sources and their integration into HDFS cluster. Hive and SQOOP were utilized for these purposes.
• In addition to existing data sources two new external data sources were identified and… Show more

Led the team and was the key technical contributor in development of Machine Learning system enabled by Hadoop to extract patterns and to help identify potential new voters, donors and to optimize funds and awareness raising campaigns.
Key Contributions:
• Supervised identification of proper data sources and their integration into HDFS cluster. Hive and SQOOP were utilized for these purposes.
• In addition to existing data sources two new external data sources were identified and integrated. Useful attributes were extracted from new sources and used for further enrichment. One of the external data sources was Twitter.
• Conducted exploratory feature analysis, followed by feature selection and ensemble learning model construction. SVM, Random Forest, and Bayesian Nets were used for these purposes, as well as Mahout and Weka. The newly created model significantly outranked existing models and control samples in real-world comparison.
Show less

During this project I led the team which constructed Data Mining and Machine Learning analytic solution to identify features and patterns in order to model and accurately predict behaviors for decision making with respect to medical claim adjustment process.
Key Contributions:
• Performed analysis, Integration, and sanitation of highly multidimensional heterogeneous data from multiple sources
• Worked closely with the team of medical professionals to perform data attribution, followed… Show more

During this project I led the team which constructed Data Mining and Machine Learning analytic solution to identify features and patterns in order to model and accurately predict behaviors for decision making with respect to medical claim adjustment process.
Key Contributions:
• Performed analysis, Integration, and sanitation of highly multidimensional heterogeneous data from multiple sources
• Worked closely with the team of medical professionals to perform data attribution, followed by feature enrichment process and correlation analysis of the features.
• Work was performed on a multimode cluster utilizing Hive.
• Used Mahout and Weka to analyze the data and to construct necessary models for decision making for near-realtime processing. Show less

This project created a new technical direction for the APL based on my proposal. Once accepted, I lead a team of people by providing technical and project management expertise. The focus of the project was to develop and implement novel methods for modeling behavior of a user based on various network characteristics.
Key Contributions:
• Proposed a novel utilization of Karhunen Loeve expansion to the PCA based approach to model users and to detect suspicious/undesired behavior.
•… Show more

This project created a new technical direction for the APL based on my proposal. Once accepted, I lead a team of people by providing technical and project management expertise. The focus of the project was to develop and implement novel methods for modeling behavior of a user based on various network characteristics.
Key Contributions:
• Proposed a novel utilization of Karhunen Loeve expansion to the PCA based approach to model users and to detect suspicious/undesired behavior.
• Initiated and participated in creation of HBase and Accumulo clusters for storage of network data.
• Developed ingestors and extractors by utilizing HBase and Accumulo APIs to interact with created clusters as well as the mathematical models. Implementation of the models included MapReduce processes as well as an efficient classifier built with help of Apache Commons Math library.
• Secured a sponsor through close collaboration with project and program managers. This resulted in the engagement of two new major contracts with up to date funding summing to 1.6 million. Show less

The goal of the project was to profile, model and undestand user behavior though topic modeling of web traffic data.
Key Contributions:
• Worked on obtaining a constant data feed through development of APIs and streamlining this process.
• Created Machine Learning based models for understanding user behavior via text analytics. MALLET was utilized for the purposes of this project.
• The results of this project had great department level reception and parts of the project and its… Show more

The goal of the project was to profile, model and undestand user behavior though topic modeling of web traffic data.
Key Contributions:
• Worked on obtaining a constant data feed through development of APIs and streamlining this process.
• Created Machine Learning based models for understanding user behavior via text analytics. MALLET was utilized for the purposes of this project.
• The results of this project had great department level reception and parts of the project and its findings were incorporated in security processes of the department. Show less