During this period, existing processes were standardized towards company-wide standards.Following a company-wide initiative, the monthly fraud and abuse meeting series evolved to a document-centric format. Each month, I prepared a few pages, including consistent metrics tracked month over month, describing current issues and initiatives related to fraud and abuse. We would then meet, take written questions on the document, and record them back. This provided more focused content and allowed for more interaction and discussion.As a new dbt-based data platform was released to inernal teams, I moved our existing transformations on to the platform. Some transformations were already in a Security specific dbt project, while others were moving from ad-hoc tools in to purpose-built models in the shared dbt.A push for higher reliability prompted us to reconsider our metrics and alerting. I contributed to efforts to move more alerting and charting to feed from structured log lines through DigitalOcean's existing elasticsearch logging pipeline.

Expanded my responsibilities, introducing two regular meeting series to bring together teams interested in Fraud and Abuse at DigitalOcean. A weekly series brought together operators and engineers within the Security group to review current metrics, plan and share tactics. A monthly series with stakeholders from across the company shared actions we took during the preceding month, their impact, and our planned future actions.During this time, I continued operating our funnel friction integrations and tools. Working together with our growth marketing department, we identified opportunities to reduce fraud by applying friction to risky groups, and increase growth by lifting friction from reliable customer profiles.I started a team within Security, Security Observability and Behavioral Analytics (SOBA). We hired a second Data Scientist who focused on strategic research and development. I supported their efforts with internal knowledge of DigitalOcean's systems, and optimization advice on Jupyter and Pandas. I filled the engineering management function while we identified a full-time EM internally.

Using data to improve the accuracy and efficiency of DigitalOcean's Security team. The Security team works to provide safety to DigitalOcean's internal operations, customers and the wider internet.We used internal data from operational databases, a data warehouse, Prometheus, Kentik and other sources to identify patterns with a high risk of abusive or fraudulent activity. Targeted friction in our onboarding funnel reduced fraud coming on to the platform. After onboarding, rules-based bots identified patterns in established accounts for manual review or automatic action.Supplementing internal data, we also integrated with several external tools and data sources including Stripe and Sift. I played a product ownership role for these integrations, making configuration changes, defining requirements for technical integrations, and sometimes building those integrations in go.

Built a system to collect data about the hardware owned by DigitalOcean across time and geography. Exposed the data to other teams to support automation and analyzed the data for trends and anomalies. A Spark Structured Streaming pipeline implemented in Spark fed the data to HDFS and MySQL. PrestoDB was used to query and analyze the data. It was exposed through a GraphQL API implemented in go, and a web UI written in Elm.

Started a team that inherited several existing integrations with third parties and built new integrations. Participated in the hiring process for remaining team members including Engineering Manager, Front End Developers, and Back End Developers. Negotiated the team's scope with the rest of SoundCloud.Also contributed to the engineering organization of the New York Office in general: reviewing code challenges, interviewing candidates, troubleshooting overall productivity issues, and discussing how local changes fit in to SoundCloud's overall architecture.

Brought together three disparate teams in to a single Core Clients team. Hired internally and externally to staff up the team. Ran visioning and planning exercises with the team to set a direction and describe how the team fit in to the rest of the organization. Ran weekly sessions to track progress and adjust direction in response to the needs of the iOS, web and Android engineering teams.

Working with iOS, Android and web teams to streamline development effort for feature teams. Introducing a release process for the mobile teams emphasizing decentralized responsibility and regular, predictable cycles. Contributing code for cross-cutting areas such as tracking and data access.

Delivering business applications using agile development practices. When desired and required, introducing new practices like Test-Driven Design, Continuous Integration and short iterations in to existing development practices.Focusing on .net technologies, but also delivering projects based on Java stacks.Participating in the open source, ThoughtWorks-sponsored CruiseControl.Net project. First as a contributor, later as the maintainer.

Implmented experiments in human visual attention and analyzed the results. Experiment presentation was done using C++ with MFC, C# with Managed DirectX, and Matlab. Analysis was done using Matlab, python and perl.Some publications from this period:http://www.cs.ualberta.ca/~wfb/publications/J-2006-Smilek.pdfhttp://www.springerlink.com/content/v3586w9872211476/

Worked with two others, creating a site for professors and University staff to store and track their CVs. Used the Seaside Smalltalk web framework and lightweight organizational processes. Met directly with the client weekly, if not more often, to showcase working software and discuss the next direction for the project.