Projects: Software

Hackathon: Machine Learning

Keras
Keras

Gartner announced they were going to host a company-wide hackathon to be held in May of 2019. I proposed a machine learning proof-of-concept to address the low quality of the pros and cons featured on our new reviews page. While leading the project, I thought it was important that everyone be able to pursue whatever approach they saw fit to learn the skills they desired to learn.

In the end, we all mostly got results of one kind or another. We learned a few things along the way. And everyone lost to the team we all knew we would lose to. We went on to implement our solution on the live site.

Product Collation

Product Filtering/Sorting
Product Filtering/Sorting

I created an object-oriented solution that also took advantage of some functional programming practices. It is a black-box service that takes an industry, all of it’s products, and a store as initial arguments. Filtering and sorting actions taken by the user trigger internal methods that act on the data and update the store.

Development of these features took a few weeks. We took the next few months to iron out bugs, uncommunicated or forgotten business logic, and other edge cases that we wanted to address.

Responsify Desktop

I argued against supporting two separate platforms that accomplish the same goal. I argued for making the desktop site responsive to replace and sunset the jQueryMobile application. My arguments won the day.

Bite Squad’s goal at that point was to have a good, solid, manageable software product to support their business. They hired talented, experienced developers to do that. They realized somewhere along the line afterward that the effort was not worth it.

Manual Collateral

Since the beginning of time, collateral has been input into the LOS by means of a form that integrates with NADA. Due to the limitations of the NADA bookout service, new and rare cars are generally unacceptable to our system. The solution was to allow manual collateral input. Because we started collateral with NADA, the system was already somewhat complex, though some care was taken to abstract out the bookout type. Still, our solution had to expand upon that abstraction, particularly in back-end persistence and front-end KnockoutJS field bindings.

Loan Calculator Bugs

There were somewhere between 15 and 20 bugs reported. The actual number of bugs was unclear because many of them were overlapping or interrelated. So for three weeks, I fixed bugs whack-a-mole style, fixing one, then another, then going back and re-fixing the original after coming back and finding it broken anew… over and over and over again.

Agile Velocity Training

Agile Velocity
Agile Velocity

FinanceGenius hired Agile Velocity to come to the office for a two-day training session to teach the team how to deliver functionality faster and better using agile methodologies. Together, we learned that an agile approach can help software companies like ours by giving more credence to the worker’s voices who know their particular specialty better than those who manage them.