Aug 2024 – Mar 2025 · 7 mos

Mar 2017 – Jul 2024 · 7 yrs 4 mos

Feb 2015 – Mar 2017 · 2 yrs 1 mo · Greater Seattle Area

• As a Senior Manager at Pro.com, I manage a team of extremely passionate Engineers across US and Chennai, India. I am responsible for key metrics to ensure that we have a good Demand Funnel at any given point, Job Delivery (Bottom of the Funnel dealing with conversions). My team is responsible for the Software Systems that determine the quality of an incoming lead and assess the Job Value and Job Conversion Potential automatically. This triage helps the company prioritize jobs appropriately. My team is also responsible for metrics around CAC, LTV and Conversion.
• Pro.com offers home services fulfilled by their In-house Pros as well as other subcontractors.My team owns the apps and other tools that allow intelligent job dispatch to these Pros along with facilitating time tracking and geo location.
• In addition to managing the team, I am the technical representative for external partner interactions, board meetings etc.

Apr 2013 – Jan 2015 · 1 yr 9 mos · Greater Seattle Area

• I took over managing the team responsible for developing software to facilitate Inbound Prep. Prep is a function that allows for inventory to be safely stored and shipped to customers without causing the inventory to be damaged. Prior to me taking on this team, decisions were made to package/prep the items by humans and it was arbitrary, in addition to costing Amazon millions of dollars. My team's charter was to learn from the human decisions and build an automated decision support system (using Machine Learning primarily) to determine how to prep inventory when needed. My team was a customer of the Amazon Machine Learning team and we collaborated to build complex ML models that determined every single packaging decision. The system is currently in use and saved over $20MM in the first year.
• In addition to this, my team was responsible to build an automated system that determined defects in inventory shipped by 3rd party sellers and other Amazon vendors. This system leveraged security cameras to take a picture/video of inventory as it was being received and this footage was audited (selectively) by an off-shore team of associates to verify that the prepping was done correctly and the items were not damaged. In addition to this, the system provided evidence to the sellers to drive an automated chargeback when items were found defective/damaged. This system accounted for savings over $40MM for Amazon.

Apr 2012 – Apr 2013 · 1 yr · Greater Seattle Area

• As a Senior Software Engineer (SDE III), I lead several high profile projects listed below:
• 1) Dock Level Loading: Allocating appointments to carriers at an FC based on the importance of freight and labor availability. Appointments were determined with carriers by a system that predicted/forecasted the labor requirements and the importance of inventory being shipped into Amazon.
• 2) Provenance Tracking of Inventory: Inventory sold by 3rd party FBA sellers were tracked all the way from their entry into an FC to the point of shipping to a customer. Such fine grained tracking allowed Amazon to fulfill customer needs better. Tracking of inventory was achieved by a means known as Virtual Tracking (which eliminated the need to sticker every single item shipped into an FC). Inventory with high concessions/defect rates, date lot items with expiry could be tracked and isolated in an FC very quickly. This project saved millions of dollars in labor and stickering for Amazon.
• 3) Performance Optimizations: Tools used by FC associates have to be extremely responsive in addition to dealing with a large number of unpredictable use-cases that are supported with 40-50 service calls to receive a single item. I worked on analyzing and improving the latency of these tools (~2 seconds tp99) so that Amazon could hit our Q4 goals in stocking up the warehouses.

May 2010 – Apr 2012 · 1 yr 11 mos · Greater Seattle Area

• As an SDE II in AFT, I was responsible as a lead engineer to design and develop the automated labor management system (Maestro). Maestro was designed and developed to replace human decisions around moving labor between different functions in a Fulfillment Center (large warehouses with millions of sqft). The system computed a buffer of work between different functions like Receive and Stow and allocated labor intelligently which included decisions made based on skill of the workers, distance to move and forecasting (more work could suddenly drop in at the Receive Dock). The system was used across the globe and was particularly used heavily in German Fulfillment Centers where efficiency of such decisions made a big difference to productivity. Receive and Stow functions improved their efficiency by 40% due to intelligent labor allocation.

Jan 2008 – May 2010 · 2 yrs 4 mos · Greater Seattle Area

• As an SDE II in the Interactive Listings Platform team, I designed and worked on high performance caching systems (pre-Sable) that were the backbone of the listing tools provided to 3rd party sellers on Amazon. Amazon has a few million active sellers that have a constant need to manage their inventory (which includes availability, pricing and new listings). The charter of this team was to provide a reliable platform on which sellers could manage their listings (spanning a few 10s to a few million with a guaranteed latency (tp99 of 3 seconds) as they paginated, sorted or generated reports and feeds. A good testimonial to the system I designed was when a seller with millions of listings who was able to modify their listings in minutes. I also worked on a Seller Desktop application (built on .NET) that allowed sellers to list and manage their inventory. I managed an intern during this time in the team who is now a very successful engineer at Amazon.

Jul 2005 – Jan 2008 · 2 yrs 6 mos · Greater Seattle Area

• As an SDE II in the EIS team, I worked on a wide variety of projects that served as a reliable platform for exchanging EDI messages between external partners (Procurement, Drop ship, Wireless and Library Services). I designed and implemented protocol engines for delivery of messages to Amazon (FTP, AS2 and HTTP(s)). The systems were designed to be large scale handling millions of transactions in a single day. I designed and implemented a configuration system that drove EDI transformations (chunking, splitting and assembly) before it got delivered to internal business partners. I worked on messaging gateways to recognize the incoming messages to route to transformation engine, the transformation engine (built on top of Contivo) and the final protocols to deliver messages (both inbound and outbound) during this time.

Apr 2004 – Aug 2004 · 4 mos · Greater Seattle Area

• As an intern at Amazon (and a graduate student at Stanford University), I focused on designing and developing a monitoring system for the Enterprise Integration Systems (EIS) team. The team had a few dozen backend services along with multiple databases (10+) and an outage in one of the systems caused a few dozen 'pager' alerts to the engineer on-call. My project was an attempt to manage dependencies along the chain, services, front-end stacks, databases, caches etc) so that it would be easy for the oncall engineer to get to the root cause instantly. This product evolved later to be Carnaval in Amazon.

Feb 2000 – Apr 2004 · 4 yrs 2 mos · San Francisco Bay Area

• At Yodlee, I worked on many different aspects of the Gatherer Technology (used to authenticate on behalf of customers and gather financial and other data). I worked as a lead engineer on a product called 'Storm' which could gather data from 1000s of websites very rapidly. In addition to this, I worked on custom products for customers like Citibank, Chase, Wells Fargo, Bank of America and MSN Money. I was in constant contact with customers right from the requirements phase all the way up to the deployment of the new products. I lead a group of engineers that managed the complete lifecycle of the products developed by the Professional Services organization.

Oct 1999 – Jan 2000 · 3 mos · Bangalore and Seattle WA

• I worked as a Software Engineer for Microland Ltd briefly before I went to the US to work with Junglee.com (which was being acquired by Amazon). I worked on technology that extracted information from the internet to facilitate selection for consumers in the form of comparison shopping. During this time, I specifically focused on building a product titled 'Universal Product Finder', a search engine that found similar products that was sold by multiple websites and an ontology to compare the prices and other features so that the end customer had a lot of information prior to buying the product.