• State of Health Estimation of Lithium Ion Batteries (August ’18 – ongoing)

[#machinelearning, #algorithms]

This is my current work at Ather, where we are trying to estimate the instantaneous capacity of the battery pack using data of a single charge cycle and no prior data. We looked at cell data of various conditions of cycle aging and calender aging and extracted features from the data that correlated with capacity fading. Currently we are working on fitting a good model between the features and the actual capacity. Post this, we would be scaling up these insights to the battery module with relevant considerations.

  • Data driven design of bicycle geometries (September ’18 – March ‘19)

[#datadrivendesign, #crowdsourcing]

This is a recent personal project on understanding how various parameters of bike geometry affect riding behavior. So far, I have identified three features which are either raw geometry values or derived values from the geometry table, which are useful in understanding what type of riding would the geometry be most suited for. Once there is a reliable technique for mapping geometry table to riding behavior, we could come up with a geometry table for a specified riding requirement. An interesting application for this would be to bring (highly expensive) high performance bikes’ riding behavior to a larger audience by replicating the geometry on a frame built locally using available and sustainable materials. This is a work in progress and I plan to open source the work once its completed from my end to let users contribute and improve it.


  • Article series on Sustainability (November ’18 – June ‘19)

[#sustainability, #outreach]

In collaboration with a friend who is a local wildlife conservator and runs a wildlife rehab shelter C.R.O.W. (Conservation and Rehabilitation of Wildlife) in Rajkot, Gujarat, I am currently writing a series of articles on sustainable living to raise consciousness in the public.

Starting with a general article on why it is important to talk right now and what are our worst impacts on the planet, we plan to then write about individual topics such as plastic free living, terrace gardening, chemical free living, responsible commuting, reducing light pollution, and other topics where each article would explain the cause and the effects of the issue along with suggestions of individual level actions.

We plan to write the articles in Gujarati & English and we plan to publish the articles in the local newspapers in Rajkot for better reach, and also publish them on our respective blogs.

  • Snap-to-Road Algorithm (October ’17 – July ‘18)


This was an interesting algorithm to tackle noisy GPS data for dashboard navigation on Ather Electric Scooters. We took into account error and heading from GPS along with latitude and longitude values and also took data of vehicle speed from motor and along with the knowledge of vehicle’s previous position, the algorithm predicts the vehicle’s position to the closest point on the navigation path.

When the vehicle is away from the path for a few consecutive points, the algorithm flags the user off route.

  • M.E.S.S.I., a Raspberry Pi compute module-based data acquisition system at Ather (August -17 – April -18)


M.E.S.S.I. is a DAQ that was conceptualized to give data of user behavior while riding for early algorithm development work within the Data Intelligence team at Ather, before Ather vehicles were launched. The first version of M.E.S.S.I. was based on Arduino and data was stored on an onboard SD card. The DAQ was mounted in the storage space in various scooters for collecting data of various scenarios. I coordinated building of v2 of the board which is based on Raspberry Pi compute module and has onboard GSM capabilities.  The version 2 of the board also has supports many more sensing capabilities. Based on the requirements of the Data Intelligence team and Product teams, I coordinated the project with the hardware team in consideration with the timeline and budget requirements.

M.E.S.S.I. v2 pilot PCB

[#HumanComputerInteraction, #FlexSensing, #PersonalFabrication]

small size images
Flex sensors within various shapes i) Inchworm, ii) Fish fin, iii) Bird wings, iv) Flower petals

For my master thesis work at Human Computer Interaction Research Group at University of Saarland, Germany, I designed and implemented shape customizable, inkjet printable flex sensors.

We identified various primitives within the basic sensor design, that can be varied depending on the target shape and required sensing resolution. The designs can be arranged in various geometries for deformation sensing in 1-& 2D geometries. We analysed the sensor response in static and dynamic deformation scenarios. We also illustrated several application cases for such quick, easy and low cost deformation sensors.

We presented and demoed the work at ACM TEI 2017 in Yokohama, Japan.

[#WearableTech, #Crowdfunding]

While at Lumos Design Technology, a startup focused on making wearable tech for bicyclists, we conducted a successful crowdfunding campaign on Indiegogo during March-May 2016.

The product pitched was Aster, a backpack with integrated lights and sensors to increase the visibility and safety of bicyclists on roads. The backpack has embedded lights (rear, front, indicator, profile & brake lights) that are controllable from a Bluetooth remote on the bike handlebar. The remote contains an Inertial Measurement Unit (IMU) and buttons.


I primarily worked on the algorithms for braking, crash and theft detection. Apart from these, I also supervised firmware and app development for Aster. We designed, developed, manufactured, tested and delivered the product to a thousand backers worldwide, within our five-member team in eighteen months.

  • System and method of contactless braking detection for a vehicle(India patent no. 300034, Grant Date: August 17, 2018)


We developed a feature based algorithm for contactless braking detection in Aster, using IMU mounted on the handlebar remote. The detection technique works across low to high braking intensity, any mounting angle of the sensor, across various positions on the vehicle frame, and across various vehicle types. We achieved superior detection timing while eliminating most instances of false detections. We filed a patent for the technique, which is currently under examination.

Aster brake patent diagram 2-01
Sequence of events: 1) Brake applied 2) IMU captures deceleration & sends brake signal to the backpack wirelessly 3) Brake lights on the bag light up
Electric Scooter data final
High intensity braking applied on a 60kmph riding speed of an electric scooter, i) Acceleration in the riding direction, ii) Ground truth for braking captured using pressure sensor on brake pads iii) Values of two features across the riding event iv) Braking detection signal generated through combination of values of both features across multiple windows

[#BrainComputerInterface, #ExperimentDesign, #GameDesign]

As part of our undergraduate thesis work, my team and I designed a platform for learning math and language. The goal of the work was to motivate math and language learning by incorporating user feedback and game based learning, especially for children who find these difficult through conventional teaching methods. The feedback was taken through EEG sensors to monitor parameters such as attention. Based on the feedback parameters, different levels of the game would be displayed for the player. As  questions to be solved are embedded within the game, the player would be keen to solve those to continue the game. We built the basic gaming platform in Scratch.

To cross the boulder, answer the following please!

While the scope of the work was too large to be completed within our undergrad thesis, it was a wonderful learning experience for us to delve into interdisciplinary problem solving interacting with students, educators, medical doctors, neuroscientists to understand their perspectives and gaining insights. It was our first attempt at conducting research through designing experiments, collecting data for various scenarios and analysing it.

[#InternetofThings, #ProximitybasedCommunication]

We developed various applications for proximity based wireless communication during EMECS-thon, a hackathon organized at University of Kaiserslautern, Germany. We presented i) a voting system for classrooms for instant feedback and ii) interacting with various sensors at home, through Bluetooth. For remote monitoring for home scenario, we also developed a web application and integrated text messaging for notifications.

Our team won the prize for Best Implementation in the IoT category among the teams at University of Kaiserslautern.

  • Embedded for Her (10-week intensive embedded systems course: June – August 2017)


I was an instructor for two sessions for the program, Embedded for Her. It was an in-depth course in embedded systems for women organized by Free Software Movement Karnataka, a non-profit organization. It was conducted with the primary objectives of developing participants’ skill set in embedded systems and increasing women participation in the embedded systems community. It consisted 5 sessions of course work and 5 sessions of project work. All the tools used and the designs developed were Free and Open Source (FOSS).

Participants tinkering and coding along during one of the sessions


Started in September 2014 by my mum and me is a web platform for sharing recipes in multiple languages.  We started the work with two objectives, i) to have web content in multiple languages (esp for those that have very little web presence), ii) to share healthy home recipes from various parts of the world to the rest of the world. Currently we have about 70 odd recipes on the platform in Gujarati (my mother tongue) written by mum and English translations of those by me. We would soon build the platform to enable other users to add content in their languages, and also translate the existing content in other languages.

If you can read Gujarati, then recipe of Bhakhri from our platform. If not, then read on the English translation.

સામગ્રી: ૨૫૦ ગ્રામ  ઘઉં નો જાડો લોટ, ૫૦ ગ્રામ ચણાનો લોટ,મેથીની ભાજી,૪ નંગ લીલા મરચા,૫ થી ૬ ફોદીના ના પાન, ૧/૨ ચમચી મરી પાઉડર, ૧/૨ ચમચી હિંગ, તેલ, સ્વાદ અનુસાર મીઠું, ઘી

બનાવવાની રીત: મેથીના પાનને ,ફોદીના ને અને મરચાને ધોઈને બારીક સુધારી લ્યો.  એક પહોળા વાસણમાં ઘઉંનો અને ચણાનો લોટ મિલાવો. તેમાં બારીક સમારેલી ઉપરની વસ્તુઓ નાંખી એકદમ મિલાવો. હવે તેમાં હિંગ, મરી પાઉડર ,મીઠું અને ૪ ચમચી તેલ નાંખી મિલાવો. થોડું..થોડું  પાણી નાંખી લોટને બાંધો. લોટ કઠણ બાંધવાનો છે. હવે તાવડી અથવા લોઢી ગરમ કરવા મુકો. મોટું લુંવું લઇ હાથ થી લોટને કુણવી ભાખરી વણી લ્યો. થોડી જાડી રાખવી.    ગરમ લોઢી પર ધીમા તાપે તેને શેકી લ્યો. બંને બાજુ સરસ ભાત પડી જાય એટલે નીચે ઉતારી લ્યો. તેના પર સરખું ઘી લગાવો. ગરમ ગરમ ભાખરી અથાણા સાથે ગોળ સાથે ખાવ.


(મારે  ત્યાં તાવડી હોય છે એટલે હું તાવડીમાં ભાખરી બનાવું છું .તાવડી એટલે માટીમાંથી બનાવેલુ વાસણ જેમાં રોટલી ,રોટલા ,ભાખરીબનાવી શકાય.) આમ તો  આ ભાખરી જ છે પણ આ ભાખરી. આ લીલીવસ્તુઓથી અદભુત સ્વાદિષ્ટ લાગે છે. આ ભાખરી  પાલક નાંખી પણ બનાવી શકાય છે.

કોઈ ને મોળા વ્રત હોય તો આ વાનગી દહીં કે શીખંડ ખાવાની બહુ મજા આવે છે!


We participated in a swarm robotics competition as part of Techfest at IIT, Bombay. The competition involved building 2 autonomous robots for navigating a mesh arena. These robots look for cube objects and deposit them at specified locations alternately. The robots emulated biological ants for cooperative task of finding food within an area.

Our bot design was highly minimal. Our team was among the national finalists for the event.

Our design (L), other designs before the competition round
  • All terrain vehicle design competition (Autofest 2007)


During our first year of undergraduate studies, we participated in an all-terrain vehicle design challenge where we had to build a remote controlled car that could traverse through different types of terrains within the arena. This event is memorable as it was the first event that we participated in our undergrad and also our first exercise towards building something. Our team also stood 3rd within 137 other teams.



Along with a few friends, I conducted an introductory workshop on Arduino. The goal of the workshop was to get kids hands-on experience on building prototypes using Arduino and realize different ideas into physical prototypes. Kids came up with novel application ideas and it was fun building prototyping and understanding basics of electronics prototyping.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.