This past summer I worked with the Touch and Sensing Team at Apple, as a Mechanical Product Design Intern.
As my main project, I led two design investigations of abnormalities seen post-manufacturing. To do this, I interfaced personally with international vendors, requesting and analyzing the results of several rounds of DOES. To further understand the problem, I also travelled to the international site, conducting an on-site process investigation and observing testing in progress. Based on my findings, I recommended a process change at 2 vendor locations that was implemented, and shown to successfully improve component reliability validation. Through this experience I was able to get an in depth look into all the parameters that come into play while taking a consumer product from concept to mass production.
Separately, I worked on an independent intern project, for which I designed and constructed a LabView driven electromechanical rig (DAQ) to automate the mechanical characterization of flex and touch sensors based on observed changes in resistance waveforms. I drafted and distributed docs for long-term internal use.
Finally, I worked closely with the team and vendors to gather necessary information to compile into engineering specification reports and technical drawings.
Due to confidentiality agreements I’m unable to post any pictures of my work. Please reach out if you would like to know more about my experience at a high-level!
I interned at Blue River Technology for 12 weeks in the summer of 2017 as a Mechanical Engineering Intern on the R&D team. Blue River Technology is an agricultural robotics startup that was acquired by John Deere for $300 million in Fall 2017. Blue Rive develops smart ‘see and spray’ tractor modules that, using computer vision, differentiate real-time between weeds and crops, and selectively spray only weeds with pesticides.
During my internship I worked with the R&D team to better automate the module’s spray process. I did this by evaluating the effectiveness of different nozzle designs, as well as by working with the algorithms team to pinpoint the time delay of the electro-mechanical system. For both of these tasks, I built an electromechanical rig for our Sunnyvale office that could mimic the activity of our huge tractor-sized modules at a smaller scale.
To evaluate various nozzle designs that varied parameters such as material selection, orifice manufacturing processes, and size I built an electromechanical rig that could mimic the activity of our tractor-sized modules at our Sunnyvale office. The electromechanical rig is pictured below. See below for more details on how I compared nozzle designs and how I investigated the system time delay.
I designed WedgeWalk after a trip to visit my grandmother in Mumbai. After returning, I kept thinking about the way my grandmother struggled with using her clunky quad-point cane, in a way that my other grandmother in LA, who used the very same cane, did not. I quickly realized that the source of the difficulty was Mumbai’s unique environment and culture. Crumbling sidewalks, non uniform staircases, compact taxis, dense crowds, and aggressive weather were just a few of the differences in terrain. Furthermore, cultural norms meant my grandmother in Mumbai more frequently donned elaborate clothing, removed her shoes, and washed her feet. The quad cane, designed for the average western lifestyle, not only failed to provide safety in these use cases, but often created more dangerous situations for my grandmother. I set out to re-design the cane.
I conducted user research, defined primary design criteria, kept a journal to sketch a cane a day, and prototyped different designs. I showcased my final design at the TechShop booth at the Bay Area Maker Faire, and was interviewed by Mike Rowe (Dirty Jobs/How it’s Made) along with the founders of Boosted Boards for CNN’s “Somebody’s Gotta Do it”. I patented WedgeWalk and am currently weighing next steps for the project.
Scroll down for more project details!
Intuitive Surgical's Da Vinci robot allows surgeons to conduct surgery through teleoperation, by providing rich visual feedback. However, the addition of tactile feedback could facilitate the development of more robust control systems for autonomous surgery. In this research project, I set out to explore the extent to which low-cost, low-resolution sensors could accurately characterize critical surgical parameters such as needle contact, needle insertion, needle deflection within a phantom, and relative angle of insertion between the needle and phantom. These parameters are evaluated with a dual capacitive-force sensor placed within a 3D printed module that fastens to the tip of the surgical instrument. The unit cost of this module is less than $10 USD. The low-resolution data set produced by these sensors proved sufficient to detect and characterize the kinematics of these common surgical procedures.
Limbr is a suite of smart exercise braces that pair with a Labview GUI to provide a user with real time feedback during their exercise routine. I worked on this project along with three other group members for UC Berkeley’s upper division ME elective, ‘Design of Microprocessor Systems’. Our smart braces consist of stretch-resistive fabric that consistently monitor a users position during the execution of user-selected exercises. Our GUI allows a user to create their own list of exercises and calibrate an ideal form for each exercise repetition. Once the user begins their exercise routine, Limbr’s smart braces and GUI work together to record user form, and display real-time whether each repetition was executed with good form. With this immediate feedback, Limbr users can focus on building good form, reduce the potential for injury, and track their progress in reaching their fitness goals.
I grew up in the Bay Area, but both sets of my grandparents were remote - Mumbai and LA. Keeping my grandparents on their respective pill schedules was a daily struggle for my parents. Complicated pill schedules are a universal problem when pills must be taken multiple times a day at specific times. I set out to understand how caregivers might best keep seniors adhered to their pill schedule.
I conducted extensive user research, ideated around the question, and began prototyping. My final design consisted of a smart pillbox that could sense when pills were not removed from a compartment, and could subsequently notify not only seniors, but most importantly, caregivers. The PillBuddy is a friendly interface that keeps both seniors and caregivers accountable.
I first internship was with D-Rev, a San Francisco nonprofit dedicated to medical device design for those living under $4/day, in Summer 2014. D-Rev served as my first introduction to how engineering and human centered design together could have a direct, positive impact on local communities.
At D-Rev I worked on a variety of projects such as prototype assembly for D-Rev’s prosthetic knee, PLM structure reorganization, and impact assessment for D-Rev’s jaundice care unit using data gathered from international partners. My two main projects however were 1) Market research for D-Rev’s $80 prosthetic knee, and 2) Quality assurance for D-Rev’s jaundice care unit. Both these projects are described in further detail below. View the full market research report here!
My first research position, as a freshman, was with the UC Berkeley Mechatronics Lab, from Spring 2016 - May 2017. Here, I gained a first understanding of satellite components, mechanics, and specifically, CubeSat design. As part of a larger research project, I designed, CAD rendered, and machined the 6U CubeSat shown below. The CubeSat included the frame, four bar boom platform, boom motor mounts, and flywheel mounts. I designed the components in Autodesk Inventor, and machined the parts at the student shop. CAD Models and the final assembly can be seen below! Watch a video of the final CubeSat in motion here.
WaldoWatch was my very first microelectronics project (Arduino) that I built in high school. WaldoWatch consists of a locator module, worn on the user’s wrist, and a receiver module, which is always attached to the items a user commonly misplaces.
Each module had its own microprocessor. The locator and receivers communicated through RF signals. The prototypes and final versions are shown below. A year after WedgeWalk, I showcased this project at the Maker Faire at the TechShop booth as well.
I built Productivity Pen with two friends over two days at Calhacks 2016. Deep in midterm season, we decided that increasing productivity was the problem we wanted to tackle. We brainstormed potential solutions and eventually, based on the amount of time we had and the parts we had readily available, we decided to create a smart pen topper. The topper held a start button, LED, and accelerometer which recorded user movement. Through calibration and testing we programmed our Arduino to characterize motions as either working, fidgeting, or stalling. We’ve open-sourced our code here. See below for project pictures!
Donec et neque sit amet nisl viverra lacinia. Aenean eu lectus purus. Donec scelerisque purus diam, id pretium magna fermentum hendrerit. Proin hendrerit erat mauris, quis rhoncus nibh lacinia quis. Curabitur orci sem, porta et eros et, rutrum tristique eros.
Vestibulum ornare nulla et turpis volutpat, rhoncus lacinia elit auctor. Quisque luctus purus neque, quis tristique felis venenatis vitae. Nam lobortis orci at erat bibendum, at fermentum ex tincidunt. Donec efficitur ligula at eros maximus mollis.
Cras sed scelerisque purus. Nullam ligula felis, tincidunt interdum bibendum at, rutrum non elit. Quisque cursus pellentesque augue id consectetur. Mauris auctor, turpis vitae porttitor sagittis, orci purus vulputate neque, ut pretium dolor enim eu velit. Quisque hendrerit est vel felis aliquet, sed porttitor turpis congue.
Morbi quis nunc volutpat, lacinia mauris eu, pretium est. Praesent eget venenatis metus, tincidunt gravida mauris. Duis eget quam aliquet, hendrerit tellus sed, egestas turpis. Morbi egestas condimentum egestas. Morbi fermentum mi ut turpis rhoncus sollicitudin. Aliquam pellentesque condimentum ullamcorper posuere nisl.