The electricity consumption of a building is a function of the number of occupants in the building and also who the occupants are. This was the hypothesis we were looking to prove. The first stage of this was then, to count the number of entries and exits a building had and also to know accurately who the occupants were without obstructing the natural activity of the occupants (hence, no RF scanner or biometrics).

Our solution, was to build a smart door frame that used skeletal parameters, gait and weight to accurately identify the people walking through.

Once the door had enough information, we used machine learning techniques to associate electricity behavior to occupants and then predict/forecast load in the future.

Many aspects of the door have been filed for patents. The research paper can be found in the publications section.

In a large building with multiple devices running at any given point in time, it is highly useful to know how much power a device is drawing at an instant. It is even more useful to know if a certain anomalous device exists in the mix, which might cause a larger power draw and thereby affect the peak power consumption.

We developed a technique of machine learning to undersatand power profiles of devices and to identify devices that did not fit the patterns. The details can be found in the publications section.

Imagine you wanted to download the latest Taylor Swift song but your phone does not have a fast enough internet connection. Wouldn't it be cool if, as you go about your day, your phone automatically connected to other peers in the vicinity to download parts of the song?

We developed an Android application that allowed your phone to switch between being a Wi-Fi hotspot, which meant you could transmit data to other users, and being a receiver which could accept data packets from other users. As a peer in the network, you exchanged information over the P2P cloud without using the internet.

For the Microcomputer Project Lab (6.115) at MIT, I built a wristband that you could wear to control characters in the video game.

This was targeted towards First Person Shooter (FPS) games where it would be fun to use gestures to shoot, draw weapons and call on help from team mates. I used a Bluetooth connection to transfer commands from the band to the computer. The sensors used were accelerometers and muscle pressure sensors. It was like the Myo armband, but before it's time.

For the Software Studio (6.170) class at MIT, we had to build a website for the course project. We built a social networking platform to facilitate community cooking.

A lot of dorms in colleges have common kitchens and a lot of students. Somedays you are too busy with homework and problem sets that you really wished someone else could cook for you. Our solution was a social network where students would volunteer to cook, assist or attend the meal. It also made sure that when you ate something, you remembered to share costs (transfer money) and to cook for everyone else sometime in the future. It also had a bunch of recipies that you could use to prepare your meal or you could add new recipies and share it with your friends.

For our Mobile Autonomous Systems Lab class at MIT, we had to build an autonomous mission completing robot. The robot was meant to map its environment and based on the results perform certain tasks within a specified period of time.

For my Embedded Systems (CS 431) class at Illinois, I wrote firmware onto a Tennsy3.0 microcontroller to emulate the famous BadUSB attack.

Traditionally, most attacks using a USB involved storing malicious files onto the storage device. When the unsuspecting user ran the program, his computer would be affected. However, BadUSB showed that you could hack into the USB handshake protocol to attack the computer without the user even having to run the malicious program.

I recreated the attack on a Windows 7 machine in order to get a remote terminal with root access. When the user connected the USB stick into the computer, the USB handshake protocol (which, I customized) would not only do its original task, but also run the malicious code I wrote onto it, thereby giving me root access to the machine as long as I was on the same network.

During the Summer and Fall semesters of 2015, I am a Research and Development Intern at Grainger's Innovation Lab in Chicago,IL. I am also the Lead Firmaware Engineer.

Grainger is the largest Business to Business Maintenance, Repair and Operations (MRO) company in the world. Which basically means that we have big corporations as our clients. What that also means is that when our clients buy products from us, they hope that we provide them with the best service.

My team at Grainger is working on building out smart connected products that can analyze and predict failures thereby allowing us to provide the highest level of service to our customers. Since we are a small team, we handle multiple projects and work on multiple levels of the technology stack wether it be wrting custom machine learning algorithms or designing distributed computing clusters to store our customer's data or writing firmware that runs on low-power, low-computation microcontrollers. We are also defining some of the Internet of Things standards and frameworks that need to be followed in building out the industrial internet.

I have a great team of mentors at Grainger, who trust my intuition and give me the freedom to try out new things while guiding me and keeping me focussed on delivering a perfect product to the customer. I can proudly say that I have lived through the whole product development cycle while at Grainger, having designed the multiple prototypes and to have finally made a market ready product.

After returning from MIT and having been bit by the research bug, I decided to spend a year exploring some of the world-class research happening at Indian universities. I was very fortunate to have met Prof. Krithi Ramamritham, who has changed my life in a lot of ways.

We, at the Smart Energy Informatics Lab, were one of the first people to be working on developing a smart power grid for India. I specifically worked on two major research projects, both related to smart buildings.

The first project was to find anomalous devices from a large mix of devices. When you are in a building with multiple devices, a single malfunctioning device could affect the power consumption by a lot, thereby leading to high electricity costs. We used machine learning tools to identify erratic appliance behavior.

The second project was to use building occupancy data to forecast power consumtion profiles for future instances of time. As a utility company, if you receive the instantaneous power requirements for a certain locality before-hand, the chances of black-outs or electricity wastage become significantly lower. We built a smart door that identified building occupants and on the basis of their past electricity consumption profiles predicted the future usage.

I met some of the smartest people I know at IITB and when we were not busy coding or designing systems we had deep philosophical discussions about life, religion and science.

A couple of my friends and I started a company when we were at MIT. This was my first experience starting a company and running it.

Having seen room-mate horror stories, we decided to build out a room-mate matcher. It was meant to be like a dating site but to find your next perfect house partner.

We had a lot of fun and learnt a lot while building the website, interfacing with customers and trying to sell the product. What I learnt most importantly though was that unless you are ready to put in a 100% into your company, don't do a startup.

During the Fall semester of my stint at MIT, I decided to explore academic research and I was lucky enough to get a chance to start my journey from one of the greatest labs in the nation.

Imagine you wanted to download the latest Taylor Swift song but your phone does not have a fast enough internet connection. Wouldn't it be cool if, as you go about your day, your phone automatically connected to other peers in the vicinity to download parts of the song? Our research involved developing such a peer to peer system that allowed for seamless and safe data transfer.

I fell in love with research after this experience. The freedom and challenges that academic research presents is what makes me go to work everyday.

During the winter of 2011, I was in the Silicon Valley of India: Bangalore. I worked as an Intern in the Wireless Research Group.

3G mobile communications had just launched in India and due to the government restrictions on bandwidth, many service providers (like Vodafone and Bharti Airtel) had complaints from customers about lower than promised mobile internet browsing speeds.

As a solution, our team was developing a Wi-Fi offload solution that allowed for seamless transition between Wi-Fi and 3G when a user was in the area of a trusted wireless hotspot. This allowed for reduction in the total number of users that were accessing the network at any given time.

I spent a lot of my downtime exploring the culture of the metropolis. The youth and liveliness of the city coupled with its amazing weather makes it a great place for a young techie.

In the Summer of 2011, I interned at the HCL assembly factory and R&D center in Puducherry. I worked on two projects during my time there.

The first one involved understanding and improving the automated computer assembly control system. All the components for the computer were manufactured in the China plant and were shipped to Puducherry to be assembled. The assembly was based on customer orders and specifications. Assembling generic computers was easy and fast. However, custom orders took longer and increased latency of the plant. I had to provide an assesment of the problem and suggest improvements to reduce overall assembly time.

The second project involved customizing the Android Honeycomb OS for the company's ME tablet series. This was my first time writing device drivers and coding in C. I guess that was what got me interested in computer systems.

When I was not at work, I was exploring the beautiful beaches and the French history of the town. It has to be one of the most beautiful places in India.

The paper shows how smart meters can be put to work, beyond their traditional use, towards explaining abnormal and unexpected patterns in the behavior of electrical appliances. We describe two kinds of uses, by visualizing and analyzing smart meter data for an academic building. The first gives us the ability to find anomalous appliance behaviors and the other helps to discover sub-optimal energy consumption patterns. The goal is to stress the fact that the ubiquitous smart meter can act as a sensor not only for monitoring consumption but also for detecting interesting energy usage patterns.

Understanding occupant-building interactions helps in personalized energy and comfort management. However, occupant identification using affordable infrastructure, remains unresolved. Our analysis of existing solutions revealed that for a building to have real-time view of occupancy state and use it intelligently, there needs to be a smart fusion of affordable, not-necessarily-smart, yet accurate enough sensors. Such a sensor fusion should aim for minimalistic user intervention while providing accurate building occupancy data. We describe an occupant detection system that accurately monitors the occupants’ count and identities in a shared office space, which can be scaled up for a building. Incorporating aspects from data analytics and sensor fusion with intuition, we have built a Smart-Door using inexpensive sensors to tackle this problem. It is a scalable, plug-and-play software architecture for flexibly realizing smart-doors using different sensors to monitor buildings with varied occupancy profiles. Further, we show various smart-energy applications of this occupancy information: detecting anomalous device behaviour and load forecasting of plug-level loads.