I am working on building products that help providers and patients. While it may seem that adding tech to medicine makes sense, it mostly gets in the way, from extremely time-sucking note-taking for providers (trust me I know, I live with one!), to filing out the same endless forms to see any provider (how many times have you had that insurance card scanned?) as a patient. How do we build products that support our medical needs as patients and providers instead of make it harder for everyone involved?

I work on AI tools to help Dungeon Master's and players enjoy their campaigns more.

I led a large, globally distributed team within the Job Seeker group at Indeed, including managers, data scientists, and machine learning engineers. I helped grow the team, hiring leaders in deep learning, senior managers, senior ICs, and created the AI/ML interview process at Indeed. The team spanned across Japan, India, Singapore, Seattle, San Francisco, Austin, and had a large remote group. The team, included multiple managers. I led ML/AI processes, including promotion, role transition, interviewing, and leveling and promotion rubrics in addition to providing technical expertise on machine learning. I led a company wide ML Platform group, including Scientists from across the company and leaders in Engineering roles. We set a 2 year development plan on how to build a more robust ecosystem of ML Platforms across the company that could meet latency, flexibility, and ml ops SLOs.

I led AI/ML roles across multiple product teams at Indeed, eventually transitioning into a role that had the scope of the Job Seeker group. I led the AI/ML team across multiple offices across the globe, visiting Tokyo and Singapore to onboard new leaders. I provided technical guidance, and leadership to the team and with Product and Engineering leaders across the company. Two major teams grew substantially in this time period within my scope, including the DSP (Data science platform) team, which supports ~40% of all deployed machine learning solutions at Indeed, and the JEM (job seeker employer match) team which built applicant quality models (predict the fit between a job and a job seeker). We worked on models ranging from decision tree ensembles (xgboost, lightgbm) to more sophisticated deep learning models (fine-tuned LLMs) I gave external talks at ICLR and O'Reilly Software Architecture covering ML Platform related topics and presenting on a panel of novel machine learning solutions related to job seeker <> employer matching models.

I began managing after working as a staff level Data Scientist and took on a reporting of both Software Engineers and Data Scientists on my team. I helped multiple teammates grow their careers through promotions and technical leadership opportunities. My team size was typically ~ 4-6 during this time, eventually including managing managers. I also worked as the technical lead for multiple teams during this time, creating, staffing, hiring, and writing OKRs, JIRA process, and agile kanban based approaches. I led planning, OKR setting, and presented progress to stakeholders. My team delivered incredible impact to job seekers by reducing friction during the apply process.

I train, test, design, and implement machine learning models to analyze patterns and predict trends to help job seekers find careers. My work is primarily done in Java and Python.

I work on developing and providing business insights using large amounts of data and machine learning techniques.

I develop and support Wicket Java applications for Investment products at USAA.

I developed software on the Planning and Scheduling team.An important part of mission operations is ensuring that the instruments take the right scientific data at the right time. The Planning & Scheduling team functions as each instrument’s time-management assistant, determining where the instrument should point, when it should take data, and what types of data it should take.To keep track of each spacecraft’s schedule, the team uses specialized planning and scheduling software. The software we use the most, which we developed here at LASP, is called the Operations and Science Instrument Support Planning and Scheduling system (OASIS-PS). OASIS-PS is capable of automatically generating operations plans based on rules and constraints programmed into the software or stored in the supporting database. Operations activities can be displayed on a timeline and operators can add, modify, or delete activities from the timeline.I developed software using Java 6, Python, SQL, for Unix, and Mac OS's.