Manoj Chandra

I managed 12 developers and worked with various dependency teams across Amazon to implement numerous changes to the Amazon Buy Box ranking algorithm. The work involved heavy use of machine learning and scientific models to optimize the algorithm to pick the best seller for a specific item. The system had to be fast enough to process over 200 million transactions an hour, going through millions of inventory and related metadata to pick the best. This was a complex project and very critical to Amazon's eCommerce store.

I managed a team of eight developers to develop HomeStay, an IoT-based smart home solution that integrates with Google Home for senior living facilities. It receives data from numerous motion and door sensors installed in facilities and provides real-time data analysis using machine learning models to provide insights into the residents. Some insights provided include: door not closed, fall detection, resident has not taken medicine today, resident has not eaten yet, time resident went to bed, and many more. The system also integrates with Google Home so residents and their kids can query the data or ask for help. Finally, the system integrates with emergency services that can send an ambulance or other care as required. The MVP got total funding and was later developed into a full-fledged product.

While I worked with DXC Technology, a client came to us with a problem about people buying multiple items and then going to another store branch or the same store hours later, picking up the same things again, showing the receipt at the front door that they got when they bought goods in the previous store, and walking out with the items. They would do this several times and hit several branches costing the company millions of dollars. I proposed a solution that processes real-time video feeds of people in the store, tracking what they have bought and when, and then providing visual alerts for potential targets based on several rules. I worked on this solution with the client and my team and iterated over it a few times before getting the green light to implement it. Also, I managed eight engineers to develop this solution tried out in five stores initially. The client loved the solution and rolled it out to 140 stores across Australia.