Sep 2017 – Apr 2020 · 2 yrs 7 mos · San Francisco Bay Area

• Leading team focused on Machine Learning products in the domain of customer service automation.
• Successfully built and delivered products in the areas of (1) automatic issue triaging, (2) automatic suggestions from knowledge base, (3) automatic agent reply generation.
• My responsibilities include:
• Working with Data Scientists to ensure that we are building the right things, being an effective representative for them for product management, other engineering teams and leadership, helping them get unblocked on data requests, helping with engineering assistance.
• Working with Machine Learning Engineers to ensure they are learning from data scientists and delivering engineering around machine learning
• Acting Product Manager for new product prototypes
• Filing Patents for new products
• Working with Product Managers that we are building the right product, with the right scoping and expectations
• Working with peer Engineering Managers to align our expectations and styles of functioning as well as resourcing for projects
• Working with Customer Success team to ensure customers are seeing results with our AI products
• Architecture discussions
• Regular 1:1s with team members
• Code reviews
• Writing code and shipping features, to support the team

Mar 2015 – Sep 2017 · 2 yrs 6 mos · San Francisco Bay Area

• Connected Car platform - plug a device into your car and the data inside the car is now exposed as an API, events (webhooks) and integrations (IFTTT, Alexa) - www.automatic.com and https://developer.automatic.com
• First Backend Engineer. Joined as 6th employee, grew to 70+ employees.
• Wrote server-side complement to new hardware with 3G modem (www.automatic.com/pro/) using TCP and protocol buffers. Started with one hardware engineer and one backend engineer (me). One year later, team grew to a dedicated product manager, 2 hardware engineers, 2 backend engineers. Became foundation of all new products.
• Wrote IFTTT integration server for Automatic Pro from design to implementation to launch in just 2 weeks. Was one of the most-talked-about features in the media articles.
• Currently working on Auto Insurance-related projects.
• Automatic Labs was acquired by Sirius XM! ( https://www.swaroopch.com/2017/04/27/automatic-acquired/ )

Feb 2014 – Dec 2014 · 10 mos · Pune/Pimpri-Chinchwad Area

• Mobile CRM (like Zendesk, but as a chat UI) - www.helpshift.com
• Led cross-functional team of 12 engineers to implement comprehensive audit trail feature (http://bit.ly/helpaudit) across entire product, requested by enterprise customers
• Responsibility of Elasticsearch clusters and performance; which were the backbone of the company.
• Prototyped efficient data synchronization using Merkle trees between mobile app and server.
• Used Clojure.

Feb 2012 – Jan 2014 · 1 yr 11 mos · India

• First Backend Engineer. Joined as 6th employee.
• Wrote ecommerce system that handled orders worth $2+ million at launch. Built on top of Amazon, Paypal, Stripe, Shipwire APIs. Handled end-to-end from checkout to shipping to inventory management to returns and exchanges.
• Handling everything server-side: infrastructure, databases, APIs, JavaScript, internal dashboards, ecommerce, user-facing applications, etc.
• Used Python, Django, Java, Clojure, MongoDB, JavaScript, jQuery, YUI, Redis, Codahale Metrics, Librato Metrics, Heroku, IronMQ, PaperTrail logging, New Relic monitoring, various APIs such as Twilio, Amazon, Paypal, Stripe, Shipwire, etc.

Jul 2011 – Dec 2011 · 5 mos · Greater Bengaluru Area

• NextDrop is a social enterprise that works with the city water board - workers call an IVR system to indicate which water line is being opened, people in the area pay to be notified and the water board pays for access to a real-time dashboard. - www.nextdrop.org
• Joined project in pilot/prototype phase, took over tech single-handedly to production and they now have paying customers.
• Built using Django, PostgreSQL, Nginx, Kookoo (voice), Netcore (sms).
• Rewrote the entire system from scratch to make use of latest technologies and best practices, including using a formal state machine, using Celery and Redis for background jobs processing, using gunicorn and gevent as the server for performance reasons, using grappelli as the dashboard, etc.
• The highlight of the rewrite was a comprehensive test suite which tested background jobs and multiple scenarios. It helped catch many bugs in the simultaneous process of development.

Oct 2010 – Dec 2011 · 1 yr 2 mos

• Joshua was the founder of del.icio.us bookmarking service, sold to Yahoo!
• Joshua had several small ideas that he envisioned, I implemented them - both frontend and backend
• Taken from idea description to implementation : ClueDB.com, a social website to share and learn tips and tricks that people use in everyday life - think "lifehacker.com by the people" - soft-launched via HN - http://news.ycombinator.com/item?id=2224883. Built using Flask, Python, MongoDB, jQuery, jQueryUI, HTML5 Boilerplate, Twitter/OAuth integration, Fabric.
• Taken from idea description to implementation : StackParts.com - a directory of resources for parts of the stack for your web application. Uses tags in creative ways to cross-reference. Built using Tornado, Python, Fabric.
• "The other exercise was to see how well i could do on a project if i wasn't actually coding it myself. i hired someone. thankfully he is excellent - i am thrilled" -- Joshua wrote in Hacker News (http://news.ycombinator.com/item?id=1980079)

Jan 2010 – Jun 2011 · 1 yr 5 mos · Greater Bengaluru Area

• Ecommerce company - www.infibeam.com
• Given problem statement of login using existing identities, implemented social auth login on top of RPX (now Janrain Social Login)
• Integrations such as the WeRead API and GoodReads API for book reviews
• Part of the team that built the next-generation ecommerce platform for Infibeam.com. First release was for www.crossword.in, the online store for one of the largest retailers in India. Areas I looked into include the content management system (described below), money currency handling and formatting, emails generation, reports generation, memcache caching, SSL, admin roles and permissions, Git workflow.
• Led team of 3 to develop an ecommerce-specialized content management system, built using Rails 2.x. Areas I managed included "pages, slots and widgets" concepts and individual widgets such as banner, navigation, product list, carousel, deal of the day widgets, etc.
• Worked on converting the Rails-based ecommerce platform into a multi-site architecture, complete with simple one-click creation of a new ecommerce website for BuildABazaar.com - a Shopify for India, that was eventually the basis for the IPO of the company (http://bit.ly/infibeamearnings).

May 2008 – Nov 2009 · 1 yr 6 mos · Greater Bengaluru Area

• Startup to market hardware products that we made
• Built, marketed and maintained ecommerce site for the ion iPod/USB Charger (www.ion.co.in)
• Part of product requirements gathering and product specification for TrackEveryCoin personal finance management system (www.trackeverycoin.com) - data entry was using a keychain-sized hardware device, and later, mobile apps
• Built entire TrackEveryCoin application backend (using Django) - architecture, implementation, test suite
• Managed interns and service providers to build applications on top of our API, for the hardware-specific desktop uploader and mobile application respectively

Nov 2006 – Apr 2008 · 1 yr 5 mos · Greater Bengaluru Area

• Part of team that built AdvancedDataGrid (http://bit.ly/advdatagrid) - a framework for advanced data representations
• Handled Linux-isms in porting of (Eclipse-based) Flex Builder IDE
• Investigated new product ideas
• Used ActionScript3

Jun 2004 – Oct 2006 · 2 yrs 4 mos · Greater Bengaluru Area

• Yahoo! Buzz Index (http://bit.ly/yahoobuzz) - statistics and data mining on Yahoo! Search traffic
• Sole engineer for a year
• Stabilized and improved all aspects, rewrote many parts of the system
• Business became successful and project grew eventually to a team of size 8
• Part of 2-member team that worked on distributed processing as the logs size per day grew larger (years later, we found out that this pattern is called map-reduce).
• Used C++, Perl.