Scott Keller Email & Phone Number

Atlanta, Ga, Us

Currently developing solutions for several programs at CDC with Databricks and Azure ADF. These solutions include elaborate meta-data models to track and move data as it is added to other systems, including moving on-premise systems to cloud. These involve many tables and data sources. These tables followed a wide variety of designs requiring some unique methods to successfully move the data in an automated way and provide a method to track and report the status. These ‘data movement’ designs… Show more Currently developing solutions for several programs at CDC with Databricks and Azure ADF. These solutions include elaborate meta-data models to track and move data as it is added to other systems, including moving on-premise systems to cloud. These involve many tables and data sources. These tables followed a wide variety of designs requiring some unique methods to successfully move the data in an automated way and provide a method to track and report the status. These ‘data movement’ designs in ADF and Databricks were made ‘Idempotent’, if they failed for almost any reason the process can simply be rerun and it will pick-up where it left off. Also worked on IZDL, Covid Vaccination information, NDSP DMI, and a variety of other modernization efforts. This includes extensive work with Databricks, Azure Data Factory. I have been working with Python for 7 yrs. I have designed systems to pull existing complex data warehouse systems to the cloud in an automated fashion. This includes automated 'meta data' systems to track the type of data, manage the last pulled information per table, by table type. A variety of strategies to pull very large tables with no primary key in a timely fashion with strategies to handle interruptions. Several of the data systems are continuously updated 24/7. I have in addition designed and developed the strategies in ADF to handle this operation. This includes reconciliation reports to track differences in the Parquet files to the source systems.I also developed a variety of Databricks to handle several unique situations on several projects, including IZDL. Show less

Greater Atlanta Area

Primarily designing data architecture for ‘knowledge management’ systems, these include OMS, PHIN VADS, RCKMS, WHO Minimal Data Set and others, all detailed below. I have researched and developed a variety of prototype projects from the time I started at the Northrop Grumman. I have decades of experience with several languages such as, SQL, C, C#, and more recently with Python, Haskell, Java, R language, etc. I also designed large hybrid SQL Document-Store systems for over twenty… Show more Primarily designing data architecture for ‘knowledge management’ systems, these include OMS, PHIN VADS, RCKMS, WHO Minimal Data Set and others, all detailed below. I have researched and developed a variety of prototype projects from the time I started at the Northrop Grumman. I have decades of experience with several languages such as, SQL, C, C#, and more recently with Python, Haskell, Java, R language, etc. I also designed large hybrid SQL Document-Store systems for over twenty years. I have also designed a network of SQL Servers that operate as one unit, all while controlling complex management tasks across the array. This SQL Server system also managed the business rules for the project, OMS. This was done with dynamic SQL code driven by custom handwritten compilers and parsers. Details of the above projects are listed below. Show less

Atlanta

One project with WHO and CDC to design a ‘minimal data set’ to provide to less developed areas of the world. This was done after the Ebola outbreak. Several systems were evaluated, including OASIS and NEMSIS emergency operation XML systems, to be reduce to a far smaller system footprint. If reduced the right way, this would inherently provide growth patterns but require a minimum of skills and resources to use. As the system and the target nation’s infrastructure progressed there would be… Show more One project with WHO and CDC to design a ‘minimal data set’ to provide to less developed areas of the world. This was done after the Ebola outbreak. Several systems were evaluated, including OASIS and NEMSIS emergency operation XML systems, to be reduce to a far smaller system footprint. If reduced the right way, this would inherently provide growth patterns but require a minimum of skills and resources to use. As the system and the target nation’s infrastructure progressed there would be knowable paths forward. Several EOC platforms were evaluated as well, to provide tangible infrastructure systems, including open source systems such as SAHANA. Provided an overall business model of the organization for an Emergency Operations Center Project, using PALNTIR. The intention here was to provide a new outside contractor, Palantir, with a model of the business and operational data structures. One Northrop Grumman project analyzed twitter data for foodborne illnesses with IBM's Content Analytics. This involved design of metadata structures to house and make searchable in the IBM system the data extracted from Twitter. A key point here is that the precise piece of information that may be relevant to an epidemiologist may be unknowable till it is asked. How the meta data describes the who, what, when, where and why must be shaped to the end task. I provided a conversion of a SAS based accounting system to a SQL Server based system to process HIV foreign aid accounting information for CDC and the State Department. This consists of converting seventy-five pages of a SAS program, with no documentation, into a SQL database structure, and integrating it with the HIV Aid accounting for both organizations. The CDC and the State Department have been using this system for several years. Show less

Atlanta

Reportable Conditions Knowledge Management System (RCKMS). The purpose of this prototype system was to review existing systems, architecture, data structures, etc. to produce an automated description of a reportable condition that could be applied to Electronic Medical Records and through a rules engine or ontology, then decide whether a condition should be reported. My instructions were to select from several models that had been analyzed as potential fits. From having worked with one team… Show more Reportable Conditions Knowledge Management System (RCKMS). The purpose of this prototype system was to review existing systems, architecture, data structures, etc. to produce an automated description of a reportable condition that could be applied to Electronic Medical Records and through a rules engine or ontology, then decide whether a condition should be reported. My instructions were to select from several models that had been analyzed as potential fits. From having worked with one team that tried to solve this problem in the past, their experience was factored in. The model was chosen primarily because it was designed as a system of rules to identify a patient's disease condition, a much more complex set than public health would use. But more important was that it separated the ‘Rules’ from the ontological uses. Past systems were ontological in nature and tended more towards ‘expert systems’. This tended towards ‘fragility’ of the system. The data structure to support RCKMS was a hybrid SQL and XML system, with the rules being kept in XML. Several rules engines were evaluated, including Drool, OpenRules, and a Dynamic Compiler. There are a variety of growth patterns after prototype to move the project forward. The project was in conjunction with the CSTE and CDC. Several systems were also evaluated to process the rules including a system based off what can be described as a ‘dynamic compiler’. There were a variety of rules engines evaluated as well. RCKMS has since progressed to stage 4 meaningful use. This system was designed to work at the state as well as national level. Show less

Designed and developed PHIN VADS, the CDC’s scientific vocabulary system. We redesigned this to a hybrid-relational database format. The design includes consumption of an XML change model to update the complex structures modeling the Coding Systems and Value Sets of those codes. There are also several search engines. Much of the database code was generated using LINQ and functional programming techniques. There was an extensive system of ‘code generators’ and parsers, which would produce the… Show more Designed and developed PHIN VADS, the CDC’s scientific vocabulary system. We redesigned this to a hybrid-relational database format. The design includes consumption of an XML change model to update the complex structures modeling the Coding Systems and Value Sets of those codes. There are also several search engines. Much of the database code was generated using LINQ and functional programming techniques. There was an extensive system of ‘code generators’ and parsers, which would produce the code, due to the complexity of a seamless interaction between SQL and XML. This includes the complex mapping between the XML change event model and its corresponding XSD definition to constrain the XML and the database transactions that will allow it to work seamlessly to update the vocabulary structures. This is still the system of record for many of the vocabulary systems at CDC. Show less

I was part of a group that researched and developed a dynamic epidemiological questionnaire tied to known vocabulary for Xforms and other dynamically produced forms and even potential dynamically produced applications. This project was started after I developed a rudimentary ‘proof of concept’ to produce a dynamic form in C#, to make a point in an ongoing discussion. This project sought to allow subject matter experts to initially design questionnaires to be rendered as an Xform, PDF or a HL7… Show more I was part of a group that researched and developed a dynamic epidemiological questionnaire tied to known vocabulary for Xforms and other dynamically produced forms and even potential dynamically produced applications. This project was started after I developed a rudimentary ‘proof of concept’ to produce a dynamic form in C#, to make a point in an ongoing discussion. This project sought to allow subject matter experts to initially design questionnaires to be rendered as an Xform, PDF or a HL7 CDA document and seamlessly be tied to an underlying data model without the subject matter expert having to be aware of the modeling effort. This was accomplished by asking the SME for clarifications on the nature and purpose of the questions. This involves a taxonomical model of the business and vocabulary as underlying infrastructure to guide the dynamic form during development. The data model is integrally bound to this process which will allow the end user of the collected data in the form to understand both the exact concept of the individual question and its precise context within a data model. This system was driven by what today would be considered an ontology. There are concepts from 'Expert Systems' that guided this system. Show less

As technical lead/system architect I was able to address a variety of problems, these included inability to hit deadlines or deliver a coherent product. Many of the problems stemmed from personnel issues and out-of-date methodologies. The CDS Warehouse receives most of its data from messages sent from the states and counties. I had the message mapping guides moved from Excel to a Meta Data Mapping database, where complex changes to the database can be generated. I also demonstrated that many of… Show more As technical lead/system architect I was able to address a variety of problems, these included inability to hit deadlines or deliver a coherent product. Many of the problems stemmed from personnel issues and out-of-date methodologies. The CDS Warehouse receives most of its data from messages sent from the states and counties. I had the message mapping guides moved from Excel to a Meta Data Mapping database, where complex changes to the database can be generated. I also demonstrated that many of the complex message structures that had been so difficult to crack open and place in a standard relational data structure can now be kept as XML and placed in a hybrid xml-relational structure. We also I hired, at my insistence, a subject matter expert with a masters in Epidemiology as part of the business modeling team, this aided communications with the Epidemiologist using the data Show less

OMS, Outbreak Management System, was a system to track and manage the knowledge during an epedemiological investigation of disease outbreaks at the state and federal level. This system was highly successful. It was a network of SQL Servers operating independently but yet fudementally as a single unit. All data was synchronized in short order with all other SQL Servers. This type of SQL Server network has many daunting challenges. Many of the traditional problems with this operating model… Show more OMS, Outbreak Management System, was a system to track and manage the knowledge during an epedemiological investigation of disease outbreaks at the state and federal level. This system was highly successful. It was a network of SQL Servers operating independently but yet fudementally as a single unit. All data was synchronized in short order with all other SQL Servers. This type of SQL Server network has many daunting challenges. Many of the traditional problems with this operating model stem from improper modeling of the data. OMS is an entity attribute value hybrid system. A public health data model was derived, correctly, from the HL7 reference information model. OMS also contained a free-form dynamic questionnaire, that could be designed by the SME in the field, yet seamlessly being tied to the correct place in the data model, allowing ease of analysis from many different disciplines. Researched and designed networking system that tied the SQL Servers together. The SQL networking system, during setup, allows the user to feed in a server topology, authentication, database server name and the replication system sets itself up with no further user involvement. This design is based in part on principals from set and category theory. The database drove and managed the business logic, in large part driven by custom hand written compilers and parsers producing dynamic code. One core design element was the introduction of ‘Uniqueidentifiers’ as keys in each table, to help solve several of the problems that occur in a DDBMS, distributed database management system. Assisted in the analysis of problems with the previous NEDDS based reference information model of the original OMS System, because of performance and analysis limitations of the original structure. Show less

Analyzed performance of a 65 gigabyte knowledge management system for entire firm. Redesigned major parts of schema, stored procedures and triggers. Designed the next generation SQL 2000 database with xml/xslt and other advanced features. Also developed a prototype data warehouse server for reporting statistics on the knowledge management system and its firm-wide web communities. Specified and designed a fail-over cluster for the knowledge management system. Evaluated and designed a… Show more Analyzed performance of a 65 gigabyte knowledge management system for entire firm. Redesigned major parts of schema, stored procedures and triggers. Designed the next generation SQL 2000 database with xml/xslt and other advanced features. Also developed a prototype data warehouse server for reporting statistics on the knowledge management system and its firm-wide web communities. Specified and designed a fail-over cluster for the knowledge management system. Evaluated and designed a distributed partitioned view on a clustered SQL Server for future scale-out. Developed and Prototyped Architecture changes to SQL , ASP and COM to allow Japanese Kanji document contribution and description to be fully integrated to the knowledge system. This also included correct dictionary sorts for each language. Among these changes are included systems to integrate Japanese and other non Latin character languages into our knowledge ontology. Redesigned and integrated “Architected Solutions” data structure, a methodology for all consulting projects firm wide and modeled the dataflow to integrate it to the knowledge management system. This system seamlessly tagged and cataloged all work products entered from Microsoft Project during a consulting engagement and forwarded the items to the Knowledge Management system with no extra effort to the user. This early Ontology was derived from a taxonomy of all goods and services that Arthur Andersen and any company that did business with them participated in. It corresponds to today tagging of an instance of a class in ontology systems such as OWL. The size of this Ontology was 30,000 to 40,000 items. On this ontology’s nodes hung 'knowledge objects', these could be complex structures with highly varied purposes. The ontology was managed by a team of 9 data stewards in our offices which in turn managed approximately 110 other data stewards from the on line communities around the world. Show less

Weeks Corp was a 2-billion-dollar public company, a REIT, located in 9 cities, performing $3/4B dollars a year in construction. Our IT department consisted of three people, this was accomplished with a few network consultants and empowering accounting staff. The responsibilities were, SQL Server, development of SQL Server, upgrading and equipment selection, all reporting for administration and accounting. Database changes and development. Programming languages used were C++… Show more Weeks Corp was a 2-billion-dollar public company, a REIT, located in 9 cities, performing $3/4B dollars a year in construction. Our IT department consisted of three people, this was accomplished with a few network consultants and empowering accounting staff. The responsibilities were, SQL Server, development of SQL Server, upgrading and equipment selection, all reporting for administration and accounting. Database changes and development. Programming languages used were C++, Powerbuilder, Microsoft Access, Visual Basic, ASP, and Arcview. Troubleshot problems with network (Windows NT and Novell), accounting and client-server database systems for Major Corporation. Supported Novell network and Timberline Account system; diagnosed problems, providing analysis and solutions; communicated observations and findings to executive level management; made recommendations for optimization of management system; directed changes in accounting system. Designed and developed a major integration of Timberline Estmating and our accounting system to produce a Volume Forecasting Sytem.Developed a Marketing system with Arcview server. This system supplied a map on a web browser, which provided a view of a city down to the individual park, building and space plan from the AutoCAD drawing and our current database information on the space. The information was searchable for any current or future lease in space by a variety of specification from the user; presented on the same screen as the space being viewed; potential clients were provided current information on building sites, space plans and significant lease availability, Information such as, dock height, ceiling height, current tenant, most recent rate paid, etc. All information was supplied from our SQL Server. System organized thousands of AutoCAD drawings. The marketing department was provided invaluable tool for providing prospective tenants with latest information, available sites and their specifications. Show less

Part of development team, which designed and developed a system to trade mutual funds and annuities on the Internet. This was written in the C language and Unix scripting in the c, t and bourne shells. This system was developed on a Sun Spark Station. Developed the systems to interface the NASD/NSF systems (part of the New York Stock Exchange) to our own data Storage system. The orders were sent from the brokers through an encrypted email system. We also wrote the encryption and compression… Show more Part of development team, which designed and developed a system to trade mutual funds and annuities on the Internet. This was written in the C language and Unix scripting in the c, t and bourne shells. This system was developed on a Sun Spark Station. Developed the systems to interface the NASD/NSF systems (part of the New York Stock Exchange) to our own data Storage system. The orders were sent from the brokers through an encrypted email system. We also wrote the encryption and compression algorithms. The orders were then sent and confirmed via the Internet at the end of the day. This system was produced to be very reliable. There could be large financial penalties for not placing a trade in a timely fashion. Show less

Responsibilities included Program development in C, Microsoft assembler, network administration and help desk duties for accounting and other software systems. Developed a system to supplement our Macola accounting system. The system would not rollup our general ledger for our 80 company stores and 200 franchise stores. Wrote a system which would perform this function. Also developed a system to post ADP payroll information to the general ledger; this was accomplished with direct entries to the… Show more Responsibilities included Program development in C, Microsoft assembler, network administration and help desk duties for accounting and other software systems. Developed a system to supplement our Macola accounting system. The system would not rollup our general ledger for our 80 company stores and 200 franchise stores. Wrote a system which would perform this function. Also developed a system to post ADP payroll information to the general ledger; this was accomplished with direct entries to the accounting data files. Supervised the team which evaluated point of sale systems and the integration of their data to our systems. Show less

Designed a database to collect Patient information and administered another multi-million dollar data collection and storage system to store and report data collected during heart catheterization at the largest medical facility in Western North Carolina. Assigned to manage HP-1000 Mainframe which collected 10 channels of analog to digital data and reported findings to physicians; determined that data collection and utilization needed upgrades; designed separate database to collect… Show more Designed a database to collect Patient information and administered another multi-million dollar data collection and storage system to store and report data collected during heart catheterization at the largest medical facility in Western North Carolina. Assigned to manage HP-1000 Mainframe which collected 10 channels of analog to digital data and reported findings to physicians; determined that data collection and utilization needed upgrades; designed separate database to collect supplementary data of patients condition; diagnosed initial problems with front end equipment in lab; coordinated solution with Hewlett-Packard service personnel; supervised installation of new front end equipment, cabling and electrical hookup during construction of new catheterization labs. Results: Database provided historical patient information for all heart surgery, invasive procedures and non-invasive procedures. This provided an invaluable source of statistics on patient survival and procedure success rates. Also designed Insurance Auditing Reconciliation Report system for accounting department in Lotus 123. Show less