The projects I work on are customized for each particular situation, so there isn’t one standard pattern, but to help you understand the types of things that work well for me, here are some examples of projects I’ve led.
Ongoing Fraud Infrastructure
Alloy was in the process of developing their Ongoing product when I joined. As part of that effort, I created a detailed design for a new parallel evaluation engine, based on their existing single evaluation engine, that could run in Redshift, enabling thousands and even millions of Entities to be processed at once, thus shifting the weight of that processing off the core system.
Along with that, I identified several key enhancements to the overall system that introduced completely new functionality for clients, opening up new opportunities. The most significant, Published Attributes (PAs), provides the ability for clients to define custom attributes that meet their unique requirements that they can then use across all of their own decisioning workflows without requiring any intervention from Alloy. Prior to this, clients had to request new support from Alloy for any additional functionality they needed and wait for that to possibly be worked into the development cycle. Now they can define and begin to use any necessary attributes immediately.
Internally, this system provides significant value, allowing us to also define new system-defined PAs, easily making data available across our entire system, including all client workflows, which was not possible before. It allows us to use a more data-driven approach instead of programmatic, speeding up our ability to deliver new functionality in key areas. There are numerous opportunities for us to use these for our own purposes within the system as well as for clients.
Beyond Published Attributes, I initiated and designed new functionality for other parts of our system such as Rule Explainability, Enhanced Output Attributes, Enhanced Custom Lists, accessing 3rd-party services from Portfolio Evaluations, Attribute History MLAs, and so on. These features provide significant enhancements, opening up new ways for our clients to use our system in more efficient ways.
Infrastructure Redesign
Blue Apron’s main system is comprised of the Consumer and Warehouse Management Systems (WMS). As initially implemented, the two were very tightly coupled making it impossible to add completely new types of functionality into the mix. Working with 12 different teams across the company, I led the effort and designed a clean interface between the two systems, providing the flexibility to support multiple Consumer frontends and multiple WMSes, as well as multiple shipments and packages per order. The design also provided for enhanced support in the Logistics system and significantly improved the feedback mechanism from WMS to the Consumer system in order to improve customer support.
We specifically structured the rollout of the project into multiple phases and, before the initial phase was even complete, the base infrastructure changes were already able to support a new and originally unanticipated pilot program that would have otherwise been impossible, thus proving the flexibility of the design for supporting expansion and continued growth.
Master Data
In addition to the Consumer and WMS systems at Blue Apron, they have many other supporting functions throughout the supply chain with each maintaining its own version of the data it needs. I led the effort to understand and document the various Master Data used throughout the system and worked with 15 teams across the company to define the source of truth for each type, along with a system for propagating that data out to the various parts of the system that required it. Part of this effort was to ensure we had consistent terminology and definitions of the data so that we could meet the needs of each supporting function as the data was distributed. Along with this effort, we identified pain points and gaps in functionality and devised strategies to eliminate duplication of effort and provide needed consistency.
Order Management
Capsule’s order management process is very complex, handling many different exceptions along the way. I helped to bring visibility and clarity across the entire process, capturing how their tasks and state machines work together, exposing inconsistencies and recommending simplifications and enhancements to make the system easier to reason about and maintain into the future. I worked with Engineering, Product, Operations, and the full spectrum of their pharmacy teams to clearly document the system details, propose new strategies to improve the workflows, and promote a consistent understanding of the system across all the teams.