Currently building data observability platform for offline ML models, including complete upstream and downstream lineage reporting & escalation. Defined site-up metric for delivery of scoring data, including availability and signal quality, with training frequency and historical contract performance as diagnostic signals.Defined Data org’s dataset site-up metric (Uptime) as our north star and wrote a contract validation & monitoring system. Alerting system reduces noise and toil for on-call, cascading reports for teams up through executive leadership. This system monitors both more tables and views, across more fabrics (internal cloud and Azure), in more data formats, more regularly (hourly instead of daily) than a previous solution built by Engineering org.

Identified the need for, wrote, and evangelized a tool for rapidly standing up ETL pipelines. Prior state forced Data Scientists to choose between a tool that did not support basic engineering principles (code versioning, review, deployment, etc) or was 100X slower (minutes -> days) to stand up ETL pipelines. Now widely used in Data org as a replacement for the first and an alternative to the second.As a side-project, built first measurements for State-Sponsored Threat (SST) enforcement models. Successfully bridged cultural and engineering gap between ML and Data Science teams.

Stopped bad people from doing bad things

Operated in hyper-ambiguous space to strategically co-define, articulate, and implement the full stack of data flows, from high-level KPIs and counter-metrics, to building and socializing an original experimentation and measurement platform, to leveraging existing infrastructure Generate business recommendations, communicated through in-person meetings, written posts and notes, and documentation, for partner teams in three continents. Achieved business results across a diverse mix of talent from very senior engineers to contingent labeling workforce. Architected and executed a crawl/walk/run approach to graph-based identification and measurement. Leveraged parallel work in another org to accelerate my team’s progress. Built a graph-based model that disabled more abusive entities than all other autoenforcement models combined. Coined and termed a latent factor, “Threat,” the minimization of which became the focus of engineering teams. To support this, first application of large-scale embedding (100Ks of features) to advertising space, leading to 99% precision on predictions against millions of “threat” entities.

I identified hacks and security breaches in real time. I combined data science with software engineering to reverse engineer malware, threats within minutes, and predict where hackers will hit next. I modeled new Exploit and Delivery threats against Windows-as-a-Service using SENSE data collection program for Windows ATP.

I fought off Russian bots and black market criminals. I built algorithms and flighting systems that killed billions of bots, and directly attacked the dark-web businesses that supported them. I built systems for I.P. Address Reputation, wrote anti-hacking tools, and direct manipulated grey-hat and black-hat marketplaces. From writing new big data abuse detection scripts, to building reports and evangelizing corporate stakeholders, I protected the safety of hundreds of millions of daily active users.

I measured the delight Microsoft’s customers felt while using our products while identifying pain points, both before and after launch. I combined physical studies, text mining, and online telemetry to identify where we drove value and where we drove customers away. Using cognitive walk-throughs, in-depth interviews, recorded sessions, and surveys, I provided evidence for the most emotional parts of computer hardware and software.

Assistant at the University of Nebraska - Lincoln. Assisted faculty in conducting research, teaching undergraduates, and community service, as needed.

Was sole graduate student working for palmtop computing initiative. Product support, assistance, and evangelization for students, faculties, and administrators as part of a computing-everywhere initiative.

Between my Management Information System and Computer Science degrees, I worked in the state government for the Bureau of Information and Telecommunications. I supported efforts to modernize the state's road sign database for the Department of Transportation, working primarily in Visual Basic and SQL Server.

Documented and automated knowledge worker workflow, leading to very large (~100%) increase in productivity, customized customer-facing properties to meet OEM specs, created first search engine for our online properties.

Served as a Tech Fellow to the College of Natural Sciences