User Portlet

Major systems I’ve worked on, ignoring working contemporaneously as an Oracle DBA (certified through 11.2), system administrator with various flavors of Unix primarily Linux, Network Administration, systems design and development team leadership. I worked long term for some companies and changed my role several times. I also did some short-term (less than 18 months) projects such as spitting and merging some of subunits of some of the world’s largest banks during the banking crisis at the end of the first decade of the 21st century.

A large warehouse type with a huge fact table containing measurements for every few minutes at every switch, substation, large commercial connections, etc. each measurement containing location code, time, KVA, KVAR and KW from which I could calculate other data such as PF, etc. I developed my own indexing strategy for time-series data that was smaller and faster than any existing method currently available. This database had a mathematic/graphic interface which provided primarily statistical and other computational capabilities, a large number of graph capabilities and simulated changes allowing the combining of circuits, loss of circuits for system protection modeling for planning, etc. I could look at graphs of a rural area and identify dairy farming areas, vegetable growth and canning areas; or industrial vs. retail commercial just from usage factors. Various decision making and “what-if” scenarios were possible.

A system that used maps, with aerial photographs warped on to them to match the aspect (matching 5-7 positive geographic points on both). We had grid then marked over this. Primarily by using the maps and photographs but also information from the system above and occasional field trips, we could identify land use, i.e. residential farm land, manufacturing, retail commercial, etc. Using historical data we developed a reasonably accurate algorithm to predict future land use. Land use has a causative correlation with power use. Transmission lines, distribution lines, substations, etc. are extremely expensive investments. If placed where they are underutilized, then the investment produces a poor return. They must be placed for maximal utilization. Transformer cost is tied to capacity, underutilization is wasted capacity, overutilization means a blown transformer or paying a neighbor to run a sprinkler on it all summer. It was only after I left that I realized the real estate investment opportunities.

Another engineer and I, in a tempestuous working relationship, designed and built a 3-D substation CAD system. Survey data was dropped in as base, every component down to nuts and bolts were tied to a database containing every detail about the component from threading to stress capacity, metal used, vendor, part number, cost, etc. We created standard assemblies that were often used with a graphic containing all parts and a database “tree” component with all the data for the parts. They could be “opened” to make specific use changes. The designers could “walk” through the substation as they designed. When complete, you would “push a button” and get multiple blueprints as required, a parts and cost list, purchase orders, etc.

While knowing but loathing COBOL myself, I wrote a program for a friend that would parse a COBOL program and create a structure chart of the code. The developers were required to create such a chart when designing a program. They found that it was easier to write the code, generate the chart and submit it for approval. If changes were required they would modify the code, generate a new chart and submit it. When the chart was approved they had the “go-ahead” to write the already completed code. This program became the single most executed mainframe program in the company.

I went to work for an agri-business company and immediately showed them a 65K license payment they recently and needlessly made because they didn’t understand the developer software licensing procedure. They were also about to buy a 450K computer to match a large client running their software. I built them two dual-processor Intel machines with sufficient disk space to do Oracle development and also build copies of client systems for debugging. The company with the $.5 mil system could not get our accounting system to run as fast as it did on our system after we recreated everything but the hardware/os. We worked with them on this (at time and expenses) for a long time before they gave up. Our software was still best of breed thanks to their design wizard and some help from me.

I went to work in bioinformatics developing a new database strategy for storing the exponentially increasing large amount of biological sequence data and it’s equally large annotative data. I developed a method for storing large amounts of textual data using CLOB storage in a relation system that provided far more rapid access to biological sequence data than previous systems when querying the annotation (associated descriptive data) categorized on load by training categorical indexing system, using text indexes with a far richer syntax added to SQL, using a general thesaurus ("rabbit"="hare") and domain specific thesauri (equating differing taxonomies and terminology). NOSQL queries using a SQL plus other, richer query capabilities that included the use of the thesauri and the weighting of the criteria which gave. via a complex calculus, ranked results ordered by a score indicating a degree of fit for each “record” to the query criteria. A subset of the results of this query of stored data provided a feed to the sequence alignment engine that used a variety of algorithms to do a variety of tasks from sequencing the rice genome to identifying victims of the terrorist attack on 9/11/2001.