Mark Soper
Product Manager: Database and MLOps
Cambridge, MA

In 2021 I joined the product team at Cockroach Labs, initially leading the operational tooling of CockroachDB Cloud: CockroachDB Cloud API, CockroachDB Cloud Terraform Provider, CockroachDB Cloud ccloud CLI. Additionally in 2022 I took on product responsibility for the CockroachDB Cloud Serverless multi-tenant distributed SQL database.

As a Software Engineer in the Azure Machine Learning Group at Microsoft's New England Research & Development (NERD) Center in Cambridge, MA, I build cloud platform software that improves the way data scientists develop and operationalize machine learning models.

Working with Rocket Insights, I led a team of 5 software engineers in developing the backend API for a Fortune 500 media company's web/mobile application. We engineered the API to scale to millions of users globally, including China.

As a software engineer and tech lead at GetHuman I built a workflow scheduling backend and associated UI that our human and automated problem solvers use to solve customer service problems. The backend is Node.js, TypeScript, and RxJS. Front end work involved Angular, React, Redux, and React Native.

I built a web application for the vSphere Docker Volume Driver (GitHub) project at VMware. The Angular application allows administrators of Docker Containers running on the VMware platform to create and manage access to vSphere Storage Platform (VSAN, VMFS, NFS) from those containers.

Matt Zisow, Chris Howes, Philippe Bosshart, and I developed and launched expert shopping service Scratch, funded by Bessemer Venture Partners, in late 2014 and early 2015. We built iOS and web consumer apps plus internal research, curation and analytics technology using Javascript, including Angular.js, Ionic Framework, Node.js, Firebase, ElasticSearch, and PostgreSQL.

I organized a two-person team to build the initial web UI for financial services startup Finmason. We worked closely with Founder/CEO Kendrick Wakeman and the engineering team on a very aggressive timeline to build a comprehensive HTML5 single-page web app based on Angular.js with a build process involving Node.js, Bower.js, Broccoli.js, Less CSS preprocessing, etc. The application involves data visualization for portfolio risk management.

I worked with the founders of education technology startup Invite Education to translate their vision - a better college planning experience for families - into a successful product. I organized a three-person team to design and build the initial version of the product, which launched privately in February 2014 and continues to exceed sales expectations. The product makes use of the Ruby on Rails and Angular frameworks and the D3 data visualization library.

I'm developing a Javascript Web UI library for exploring W3C Open Annotation Data, working with the Harvard University's CATCH project team. The library will allow researchers to navigate and discover annotations of video, image and text media for peer review and other research activity.

I'm helping Oxfam America build and deploy their next-generation web platform for publishing and fundraising. We're using Python, Django, Mezannine CMS, Javascript, etc.

Led initial front-end development of Kickbit, an iOS/Android mobile app for early-stage startup Aquto - funded by North Bridge Venture Partners and Matrix Partners - that is pioneering mobile carrier plan data as form of virtual good/currency, making mobile data more affordable for consumers.

Working with a UI/UX designer and another developer, I designed and implemented initial build/tooling/workflow process for supporting wide range of device platforms (iOS, Android, web, etc.). I defined and implemented the initial application architecture, played a strong role in UI/UX design, and helped define/implement customer development and usability testing.

Product technology specifics are confidential. I learned a lot about, evaluated and implemented technologies in several categories: 1) Javascript client frameworks like Ember, Backbone, Angular, etc. 2) Javascript build and tooling technologies like Yeoman, Grunt, etc. 3) Hybrid native/web bridge technologies like Phonegap, Titanium, Trigger.io, etc. 4) DOM libraries like JQuery, Zepto, etc. 5) Javascript testing tools like Mocha, PhantomJS, Jasmine, Selenium, etc.

More about Aquto

The fifth employee at Connected Sports (backed by Avalon Ventures) I was the primary developer for the Javascript / HTML5 front-end for the company's first version of the Rumble Football product and built much of the supporting Node.js REST API. A social 2nd screen companion game experience for watching NFL on TV, Rumble Football involved real-time communication using Web Sockets and Socket.io, use of the Backbone front-end framework, JQuery, JQuery UI, Node.js, Express, MongoDB, and Redis.

More about Connected Sports

Contract project: Built initial prototype of HTML5 Mobile Web Experience and helped establish initial Node.js API for real-time collaboration startup funded by Google Ventures.

More about Collaborate.com

Helped build initial HTML5 UI / UX and Node.js backend for a consumer mobile web app that won the social good category at AngelHack Boston in 2011. Established the Node.js Wrapper for Runkeeper Health Graph API open source project.

More about Fitgiver

Contact project: Helped build initial Javascript SDK for Grocery Products Mobile Advertising Network Swoop (funded by US Venture Partners)

More about Swoop

I created a service called Touted.com in 2009 to aggregate Twitter conversation about local businesses to help consumers discover the best bars, restaurants, etc. nearby. The service pivoted to become Rally and focus on the long-tail of local events to help people make casual plans with friends.

Rally mined Twitter for content related to the future, extracting data about date, location, event, people involved, conversation thread, etc to build a calendar of future events in users' social graphs. The main user experience was a daily email that aided last-minute planning. While the daily email was successful among initial users (5% daily clickthrough rate), attempts to discover a workable distribution model failed, and the service was shut down in 2011.

I designed and developed all aspects of the Rally product, from email and web user experiences to Python / Django API with Twitter Streaming API integration, MongoDB, Redis, RabbitMQ, Amazon AWS, etc.

More about Rally

I founded Likematter in 2008 to help people discover relevant news and job listings on the web. Likematter built profiles of personal interests from favorited content like bookmarks, blogs (OPML), etc., crawled the web for current news and job listings, and the recommended highly relevant new content using machine learning and natural language processing based on the OpenCalais entity extraction platform.

I led a 2-4 person team developing natural language processing and machine learning algorithms in Python and a Javascript/HTML/CSS UI / UX for personalized news and jobs discovery.

More about Likematter

Collaborated with Chris Mihm to build an NCAA Basketball Tournament bracket manager that identifies "undervalued" teams by comparing Vegas odds with tournament seeds.