Personal Portfolio - jmadison222/knowledge GitHub Wiki

IEEE - Agile Architecture Interactions - It was obvious in my college years that ACM and IEEE were the two most impressive places to be published. Then IEEE had a call for papers on the topic of Agile and Architecture at the very time I was the main company trainer on Agile, and had been an architect on the Enterprise Architecture team for several years. So I wrote a paper, and it got published! Well, yeah, there was also that part about re-re-re-re-writing it only to get slammed by the peer review and re-re-re-re-write it. I try to block that part out.

CAS - Very Large Calculation Systems - I had written or been involved with a number of projects that seemed to have a pattern. It wasn’t ETL. and it wasn’t Data Science. It was like a really big calculator between those two. The main topics that I experienced were actuarial ratemaking and loss reserving, where many years of data are recalculated against current assumptions to see what it might look like. But the notion can extend to other things. I didn’t see this pattern covered anywhere else in industry literature, so I decided to write about it. It was ultimately published by the Casualty Actuarial Society, and I presented it both regionally and nationally.

• Fun story…​a few years later, I was at another actuarial conference, and some speakers were presenting things that were amazingly adjacent to what was in my paper. I went to them after their talk and mentioned my paper. They said, yes, they knew of it, and it was inspiration for their big project at their company and core to what they were presenting at the conference! It was inspiring to see my work come full circle.

Artificial Intelligence Agility for FIMA - Keynote speaker at a financial conference. More to come…​

Oracle Technical Network - Building a Hybrid Data Warehouse Model - My Master’s thesis detailed how to combine normalized and dimensional modeling into a single schema. It was 10,000 words. I mashed it down under 3,000 to get it published on OTN—​and didn’t lose any content, which tells you how bloated academic writing is! And they paid me $2,500, which I wasn’t expecting at all—​I just wanted bragging rights. The core of the design is simple—​both the notion of relational relationships and dimensional relationships are just a bunch of foreign keys. Nothing prevents putting both types of foreign keys in tables and using just the ones you need. I was able to run both types of queries on a single model and it worked great. The one downside is the optimizer would get confused sometimes and it could impact performance. But that was small compared to the advantages of a having two models in one.

BIJ - SOA Versus Service Chaos - When service oriented architecture (SOA) was at the peak of the hype curve, I was doing lots of it—​so was everyone else. I was at a happy hour and ran into a former coworker who was now also doing SOA. We both agreed that if you just publish lots of services, you don’t have an architecture. But what exactly did that mean? We didn’t have time to dig into it, but it made me think. The results of that thinking became this article. It’s the list of specific things to look for in a portfolio of services to determine if you have a planned architecture or general chaos.

VLCS - Overview v2.0 - The presentation that goes with the paper. More to come…​

CIO Magazine - 2008 Top 100 Award - I built the geospatial system for the company as part of a catastrophe management system that won an award. More to come…​

Computerworld - Best Practices in Business Intelligence Award - Our company took first place for an architecture that combined the best of IT controls and business freedom. The "Common Analytic Layer" had controlled assets from IT that provided the starting point for data science work without requiring the data scientists to do the heavy lifting needed for those large assets. But it also provided a high degree of freedom to then build further on those assets with the more nuanced techniques that would have been stifled if they were turned into requirements and put through the IT delivery process. An example of a key part of the design was the publication of flat-and-wide (F&W) tables. IT loves normalization. Data science loves F&W tables. They used to make them all themselves, but in this design, we built their core F&W tables so they didn’t have to. We then gave them database write authority so they could join more element and entities to the big F&W tables as needed.

Independent Oracle Users Group - Multi-Terabyte Database Migration - We migrated lots of big databases, so we wrote a paper on it. More to come…​

Independent Oracle Users Group - SOA Design Patterns for EDW Delivery - More to come…​

OpenWorld - Exadata Top 10 Lessons Learned - More to come…​

OpenWorld - Exadata with TCS - As part of our overall Exadata work, we partnered heavily with the vendor TCS. We defined the architecture and the overall approach, and they provided staffing for the bulk of the work. They asked us to present the partnership in their time slot at Oracle OpenWorld. The presentation included many of the same technical details as other presentations we gave, but with an emphasis on the use of external staffing with TCS as the vendor.

Oracle Technical Network - Multi-Terabyte Database Migration - Back in the days when a terabyte was a large amount of data, a very skilled DBA and I did a number of migrations of large databases. He was the brains of the operation, but I handled the application layer and organized the project, so we published the approach together on OTN.