Helping to build services to leverage LLM and in-house NLP models for chart analysis.

* Transitioned to working with foundational models like YOLO, optimizing detection systems by integrating camera calibration, statistical analysis, and geometry to improve shopper action detection.* Developed a new method for our human product identification task using a reinforcement learning approach inspired by multiarmed bandit algorithms, driving improvements in accuracy and efficiency.

* Temporarily managed the Computer Vision team, ensuring product requirements were complete and actionable, providing project planning, tracking, and offering technical and career guidance to the team.* After hiring a new director of computer vision, founded and led a small MLOps team to address deployment and retraining challenges.* Developed automated pipelines using ML Flow and custom tools for training, validating, and deploying machine learning models. Built a comprehensive test system to validate models against production data using annotated ground truth.

* Spearheaded the deployment and architectural groundwork for Grabango’s cutting-edge checkout-free technology.* Led the transition to Docker containers deployed on Google managed Kubernetes, optimizing deployment efficiency and ensuring seamless scalability.* Developed the company's first scalable queueing system for human review tasks using Celery and Redis, enabling faster and more reliable handling of review processes.Designed, implemented, and maintained a highly scalable, cloud-based Kubernetes system for running deep learning inference on thousands of videos, utilizing spot instances to reduce costs with robust failover and retry mechanisms.* Collaborated with cross-functional teams to ensure that the cloud infrastructure met both current and future scaling needs, contributing to significant cost reductions and improved system resilience.

At Jawbone Health Hub I have worked primarily on developing new micro services in Python to support our data collection and analysis. I've worked with our team to introduce Kafka and Celery to begin moving to an event based architecture where appropriate. The current system has provided me with opportunities to use Python, Flask, Google Cloud Platform, Kubernetes, Docker, Cassandra, MySQL, Redis, Celery, Kafka, and other technologies.

iCracked has a dynamic startup environment providing a wide array of opportunities. When I began at iCracked I was hired into a software developer role but soon transitioned to a leadership position to help guide the technical choices around infrastructure and software architecture. In my time there I have implemented software and performed devops tasks to keep our team moving towards the company goals. Some highlights:◆ Implemented basic network security layers by adding public and private subnets, an OpenVPN server, strict routing rules, and firewall/security group settings.◆ Replaced single points of failure with clustered, load balanced services.◆ Re-written key services using AWS Lambda (NodeJS) and API Gateway to lower costs and improve scalability.◆ Ported existing services to docker containers and built scalable infrastructure with AWS ECS. ◆ Helped configure CI systems (Jenkins) to perform automated testing and continuous deployments.◆ Added system level monitoring to persistent hosts using Icinga.◆ Helped develop new processes for our organization to improve testing and documentation.◆ Oversaw and participated in the development of RESTful APIs for partner integrations.

In my time at LiveOps (now Serenova) I was given opportunities to work in many of the systems throughout our stack. I started as a Software Developer but progressed to Software Architect. While at LiveOps I was able to work in Java, Clojure, Perl, and Javascript through a complicated system that included in house, proprietary, telephony, payment, reporting, and workforce management. Some examples:◆ Worked with a small team to develop a new generation of micro services utilizing the Netflix OS tools.◆ Implemented a new telephony call management service in Clojure using Twilio.◆ Designed and developed several components of our reporting pipeline using Kafka and Clojure.◆ Worked on a small team to develop our next generation SIP Server in Java.◆ Implemented key features including a way to serialize active calls to disk in special cases, relieving system load. ◆ Migrated several systems from CVS, to SVN, to Git.◆ Added Jenkins CI systems.◆ Designed and developed a new generation of our incentive system in Perl.◆ Built a self service interface to allow business users to define rules that had previously required a developer and SQL.◆ Improved overall payment system performance by add queues and improving SQL queries.◆ Maintained and improved our agent payment system and interface.