Barry Hart Email & Phone Number

Atlanta, Ga, Us

I am an adept Machine Learning Engineer with a strong track record in delivering robust and performant ML solutions. My consulting experience is built on a foundation of effective communication, collaboration, and extensive technical know-how.Key Expertise & Achievements:* Architected and rolled out ML infrastructure including feature stores and MLflow* ML experimentation pipeline and model deployment* Leveraged Google Cloud technologies such as Kubernetes, Airflow, Cloud Run, and App Engine* Provided guidance and support to a team of 10+ data scientists* Proactive monitoring of systems performance and stability using Datadog* Expert in Python language and ecosystem* Proud contributor to notable open-source projects, including Airflow and MLflow

Open for SQLFluff-related consulting and support.Developed the dbt and Jinja templating algorithm. Develop and maintain lint rules and core linter.

Taking time off after tumultuous Intuit acquisition of Mailchimp to think about next steps in life and career, preparing for transition to part-time work.

Atlanta, Ga, Us

Staff Engineer on ML Engineering and Platform.- Lead engineer and product owner for Mailchimp's data science platform- Reduced the time for productizing models from years down to days or weeks.- Support and mentor data scientists building and deploying models- Led data science adoption of Google Cloud, Kubernetes, BigQuery, Cloud Dataflow (Apache Beam), Airflow, Apache Hive, and Apache Spark- Led the engineering effort for a user CLV model that processes ~1 billion records daily.- Led development of a job triggered by Airflow that replaced a custom Python solution. The project involved converting 1 terabyte of data from an in-house Elasticsearch cluster to Avro, uploading to Google Cloud, processing the data in Apache Hive, then downloading the results back to in-house Elasticsearch.- Prototyped a recommender system using two different technology stacks: - MLLib's Alternating Least Squares, Apache Spark, DataProc - Distributed TensorFlow and WALSModel (Weighted Alternating Least Squares)

Atlanta, Georgia, Us

Hands-on architect and team lead with broad responsibility over development of back-end services for Pindrop Protect, an anti-fraud product which classifies incoming phone calls as fraud or not within 30 seconds of answering.Summary: At Pindrop, I consistently demonstrated the ability to produce complex components in unfamiliar areas, delivering high quality solutions along or with a small team in a few months. Worked closely with others in a variety of roles -- support, data scientists, product management, marketing, network operations, and other developers.During my time at Pindrop:-• The development team grew from 3 developers to 40•- Pindrop grew from a $30 million valuation to over $200 million- • Pindrop's sales grew over 10x, with new customers including a number of top 10 banks and brokeragesResponsible for:•- Scaling and architecture of the Protect application. On a 4 month project, increased call capture performance by 400% by introducing Cython and moving from a dual-process architecture to a fully parallel architecture based on Python multiprocessing. Developed a distributed, high-speed packet capture service in C++ using Boost, ZeroMQ, and libpcap.•- Audio conversion and streaming server used by fraud analysts to review fraudulent phone calls. Based on Twisted and Numpy.•- Lead developer on a project to integrate Protect with Avaya telephony systems. In 3 months and with no formal training or prior experience with Avaya or other telephony systems, developed an integration service in C++ using the TSAPI SDK, coordinated with the customer team to test and then deploy at multiple Pindrop customers, while simultaneously undergoing certification with Avaya.

Marietta, Georgia, Us

Architecture and design for AirSage's big data applications. One project was a population analytics application. The application served a number of our clients, processing billions of data points each week on two CentOS clusters, each with 1,000+ CPUs.In a 3-month period, working mostly alone (with part-time help from an Ops engineer and a UI engineer), we turned a set of loosely connected traffic analysis algorithms into a configurable, highly automated big data workflow. This work became the foundation of AirSage's new Population Analytics product line, the fastest-growing part of the business. In developing the final product, I developed a new big data framework based on Condor and Gluster to replace Hadoop and Redis. The new framework is more powerful and easier to use and extend.On another project, I researched Storm and developed AirSage's first topology for it. As a part of this project, I created and open-sourced the "Petrel" library (https://github.com/AirSage/Petrel). Petrel was the first tool for developing Storm applications in a non-JVM language.Used Puppet, Vagrant and VirtualBox to automate testing and deployment of these projects on CentOS 5. My advocacy of Puppet led to its widespread adoption within AirSage, resulting in greater consistency and productivity for the operations team.

Atlanta, Ga, Us

Senior developer on an SaaS application for a top 10 retailer which will enable them to run a common set of promotions across a variety of media channels. The application is deployed on Amazon EC2.My responsibilities include:- Developing a reusable set of UI components, build tools, and automated test infrastructure- Identify and evaluate new tools and techniques for making the team more productive- Coordinate with business analysts and platform teams- Interview and mentor new developers

Atlanta, Ga, Us

Lead developer for UIBlox, a declarative web framework. The framework is implemented primarily in Python and Datalog.

Developing web-based retail planning and optimization applications. Primary technology: Python, TurboGears, Ajax, and PostgreSQL

Atlanta, Ga, Us

Technical lead for the C++ portion of the Optimi's common software platform. 50% development, 50% mentoring, support, recruiting, etc.

Austin, Texas, Us

Senior developer on RPAS, the platform for the Retek division's analytical retail applications. Responsible for the core calculation engine, built-in extension modules, extension API, and overall platform performance.

Ghjkl, Sz

Madison, Al, Us