I help develop backend software at Perch, a technology-driven commerce company that acquires and operates Amazon third-party and other D2C brands at scale.I have been primarily involved with backend software designed to drive the growth of Perch’s products. I am heavily l involved in writing Python code, making use of Django, PostgreSQL, Snowflake, and S3. I’ve led the backend effort for demand-based forecasting and have also supported interfacing with Amazon’s Advertising API to support test accounts. As a lead I’ve planned sprints, prioritized with PMs, identified needed features/tech debt.

At Indigo I have been helping create our carbon credit generation. I have written software that analyzes data submitted by growers in our program. It takes this data through various stages of validation, making certain there is sufficient data to generate carbon credits, the data is reasonable, and that it supports additionally.I also help growers submit their data to Indigo more easily via software that extracts customer data from non-Indigo sources and converts it into the formats expected for carbon credit generation. This makes use of OAuth2 to access external data, proper encryption of keys to protect customer data, and provenance data to provide audit capabilities.Our tech stack includes tools such as Python, GitHub, CircleCI, AWS, PostgreSQL, DataDogSentry.I am also co-lead of Indigo disability employee resource group.

As a member of the Guest Data Platform team, I’ve helped restaurants better understand their customers - while still protecting customers’ privacy. This has included tokenizing/detokenizing personal information, understanding customer patterns, determining marketing return on investment, and complying with appropriate privacy laws such as the California Consumer Protection Act. Most of our code is in Java microservices. The technology stack includes AWS, Docker, PostgreSQL, and Spark on Zeppelin. As part of our Covid-19 Rally For Restaurants campaign, I developed a Python-based tool which gathered restaurant data from a variety of sources, normalized and de-duplicated the data, and uploaded it to Algolia. This data allowed consumers to support their local restaurants through online ordering and gift cards and included both Toast and non-Toast restaurants. This effort focused on an extremely quick first deliverable which raised several million dollars in gift card purchases in its first two weeks. The technologies used included Python, Pandas, and AWS Lambda.

I helped develop and support our data science platform for the Experiences business unit. This platform provides an infrastructure for data scientist pipelines. It enables reproducible batched training and the deployment of microservices for real-time predictions. These predictions provide an individualized customer experience.Made numerous improvements to data science platform in reliability, speed, storage, and usability.Developed feature store solution using Redis.Developed Java tool leveraging swagger-codegen to validate/test OpenAPI contracts.Productized data scientist algorithms. This involved cleaning and fine-tuning their code (typically Python, sometimes R), performance tuning, developing Rest API contract using swagger, determining needed Kubernetes resources, and resolving dependency problems. Led nearly 100 volunteers to help the Refugee Youth Summer Enrichment program.

As part of a corporate quality improvement drive, performed data analytics to identify trends on customer service requests throughout the data protection division. Used ETL techniques to combine data from a variety of sources in Greenplum Big Data Lake using SQL and Domo. Cleaned data to account for many missing data points. Determined trends, identified low-hanging fruits, worked with support and development to determine courses of action. Provided dashboards for tracking progress towards 25% reduction goal. Developed reports for deep insights to drive actions. Performed analytics for competitive intelligence. Leveraged installation information to understand customer environments to identify potential opportunities for competitor displacement.Served as "defect czar", the primary data analytics and steward role for the VPLEX organization’s defect tracking. Performed ETL operations on the defect database and reported/analyzed via Tableau dashboards. Dashboards tracked incoming/burndown rates and leveraged previous release information to predict future find and fix trends. Worked with product owners and customer support to prioritize fixes and managed the defect deferral process. Made certain data and metadata were accurate.Led cross-functional software Data Unavailability/Data Loss (DU/DL) team. Analyzed customer events, determined needed software and testing enhancements, and documented techniques for avoidance, mitigation, or recovery. Developed and reported visualizations for events. Efforts resulted in a measurable and sustained reduction in DU/DL rate.

Led team and developed software to manage back-end paths to storage arrays for Invista storage virtualization product. This included handling IO errors; sending SCSI commands; responding to SCSI unit attention; managing active/passive, active/active, and asymmetrical logical unit access (ALUA). Developed code to handle properties of different storage arrays, often by reverse engineering.Developed software to manage communication between dual storage processors on Invista. This included detecting initial heartbeat, determining master/slave relationship, avoiding split-brain scenarios, and devised mechanism to offload portions of path management to attached switches. Developed US Patent 7,818,606 to manage trespass decision making. Engineering contact for support escalations. Diagnosed critical customer issues live, often working directly with customers. Methods used included memory dumps, logs, and live analysis. Leveraged Python-based tool to collect diagnostic information. Helped set development/sustaining agenda for software enhancements. Trained support in diagnostic/recovery techniques to minimize escalations to engineering.

Pre-IPO startup. Developed embedded software for Cereva-5000 Storage System. Specifics included implementation of new RAID algorithms, failure recovery, and parity and tag generation/validation.

Developed embedded software for Cascade family of switches (i.e. B-STDX-9000, CBX-500, GX-550). Specifics included SONET/SDH framer drivers, system timing software, switch management. Technical lead for layer one software of MCS-25000 core switch. Initial company was Cascade Communications, acquired by Ascend Communications, and finally acquired by Lucent Technologies.

Developed embedded software for General DataComm's DV2 Switch. Specifics included Frame Relay/ATM Interworking, Circuit Emulation, multicast implementation.