In 2017 I decided I wanted to try developing a product I could sell in retail stores, while only putting 10 hours a week in, leveraging my skills in user research, engineering and manufacturing.
I brainstormed a list of possible products, prioritized and eventually down-selected to fidget spinners. Yes, fidget spinners - the fastest growing category of toy in decades.
Through my research I learned there was a large demand for the anxiety relievers in the "everyday carry" community (EDC). A male-dominant market of knife collectors who spend coin on well designed goods.
I prototyped a few spinning knives and collected feedback from members of the EDC community until the design was perfected. A flick knife with an AUS8 Japanese steel drop point blade, sporting ceramic bearings.
I sold 1,000+ units of a product I designed, Fidget Knife, which was distributed through brands such as Touch of Modern and Huckleberry, and gained millions of views through my account based marketing efforts.
$60 MSRP, $10 COGS, profitable!
Meteorite was popular on various Youtube channels. This one with Techsmartt yielded ~1.3M views.
I partnered with various retailers to distribute the product
May 2015 - August 2015 I worked as a contractor at Google's Advanced Technologies and Projects Group (ATAP) on the modular smartphone, Project Ara.
ATAP is a specials ops division of Google that operates like a start-up. I contributed as a mechanical prototyper and interaction designer. Since the electropermanent magnet latch that was marketed in the first release had failed, we needed to start from scratch. We had to design a latching mechanism that would provide a magical experience, while meeting robustness specs.
With the team, I brainstormed and prototyped mechanisms for module ejection and insertion. This was tricky, because we needed to minimize it's Z-profile, so as not to make the phone unduly thick or reduce battery volume. But we also needed to minimize the amount of area it occupied, to maximize the real estate available to developers.
My biggest contributions include:
Producing mock-ups, electromechanical prototypes to walk through on screen inputs and ejection, and mechanisms to help define the direction we should proceed.
Conceiving of an interaction in which users can physically tap modules to call software functions, without needing to go through mobile apps. A utility patent was filed on this.
This feature can be seen in the Project Ara promo video at 0:49 - 0:52.
Texas Medical Center, medical device start-up
Metered-dose inhalers haven't changed in 40 years. Having friends and family with asthma, I wanted to create something better. I believe the best design as transparent - out of our way until needed.
The problem with metered-dose inhalers is that they are bulky. This leads to misplacing, forgetting or intentionally not carrying the inhaler. 63% of asthmatics have had an attack with no inhaler nearby, 97% of the time due to forgetting or misplacing their inhaler.
The Bloom Inhaler is a credit-card sized rescue inhaler. It holds up to 8 doses of albuterol, and fits easily in a wallet or tight pocket. The initial concept was slated to be "refillable" using your existing pMDI canister.
Bloom is an international finalist (Top 20) in the James Dyson Design Competition.
To validate the initial hypothesis that inhalers were bulky, and often not carried as a result, I sought to answer three initial questions:
1. How often do people need their rescue inhaler?
2. How many doses do they take when they need it?
3. What don't they like about their inhaler?
First step - survey. With a $40 post on Craigslist for an "Asthma Focus Group" in NYC, I captured 500 responses from asthmatics through a "screening application" linked to Typeform.
These early surveys were insightful and continued to validate the usefulness of a discrete inhaler.
But there are different kinds of asthmatics in different demographics. Focusing on the 20-40 year old bracket, with mild-intermittent to moderate-persistent asthma, I held focus groups and conducted interviews using my Craigslist contacts.
I presented renderings and 3D printed models to interviewees for feedback. Body language and micro expressions often revealed problems with early designs. Excitement for the concept was met with awkward fumbling of the mockups. Encouraged, I continued to iterate until patients were satisfied.
Ultimately I came to a form that patients seemed happy with. The trigger was smoothed to eliminate the errors I was observing in how to actuate the device. The edges were rounded to make slipping into the sleeve of a wallet effortless. The orifice of the nozzle was placed higher up, like a breath spray, so the rest of the device could hide in your palm making it discrete. The color, material, finish was designed to evoke a premium feel so asthmatics could be proud of their inhaler, as opposed to embarrassment as usual.
I had an concept that seemed to resonate with people in face-to-face meetings. But I wanted to build a business and needed to understand how it would be received at scale. Using Adobe Creative suite, I put some animations and renders together in a clean website and sent to MedGadget.
When the story broke, it went viral. Gizmodo, CBS, and others wrote about Bloom. Posting in Facebook groups and Reddit drew a massive percentage of the website traffic, to complement the press coverage. We got 10,000 unique visitors and nearly 1,000 reservations. 10% is impressive conversation, given the e-tail national average is about 3%. This suggested to me there is a product to sell.
After this coverage in March 2016, I brought on an incredible team of engineers and advisors. We went on to interview pulmonologists, general practitioners, and other demographics of asthmatics to further refine the design. We have filed our IP on novel valve and pressure vessel designs, and created working prototypes.
We were accepted into the Texas Medical Center medical device incubator in August, and were humbled to be selected as international top 20 finalists for the James Dyson Award.
After speaking with more clinicians, we learned that the biggest problem with pressure metered dose inhalers is hand-breath coordination. Often patients will dispense a dose before inhaling, or far too late when the lungs are already full of air. Clinicians see patients making all kinds of errors - such as failing to shake their device, or forgetting to hold their breath for at least 10 seconds after inhaling. They were skeptical of an inhaler without a mouthpiece, as it creates yet another failure mode: poor concentricity of the spray head and the oropharynx.
Taking these concerns into account we've designed a collapsable mouthpiece into our inhaler.
We've also changed the way in which patients interact with Bloom. There is a strong culture associated with using puffers, so we redesigned Bloom using the puffer as an interface metaphor. The inhaler is actuated by pressing the top of the device, while the patient's lips are securely fixed around the mouthpiece. We account for false positives by locking the device with a slider.
Article in the New York Times about our work on Level 3 autonomy.
https://www.nytimes.com/2017/06/07/technology/google-self-driving-cars-handoff-problem.html
I was featured sitting in the simulator I built in the article
Kinesis is a smart shirt that serves as your personal trainer.
Most fitness wearables today are great at giving you data like steps and heart rate, but they don't tell you what to make of that data.
Kinesis tells you in real-time if your form is off to prevent injury and help you get a better workout.
9 Axis IMU, Skeletal Tracking proof of concept, real time rendering in Blender
If wearing headphones, the wearer can be told to adjust their form
This 2x2 describes gym goers. Synthesizing 100+ interviews. Kinesis would best serve self-motivated individuals new to exercise, who need to teach themselves.
Most people are introduced to exercise by a friend. Those who do not have that luxury rely on magazines and YouTube videos. These sources of information are limited as they do not correct your mistakes. This is where Kinesis fits best.
The Cycle Tester was designed and engineered for OXO Labs. OXO needed a cycle tester able to determine failure modes and cycles to failure. The machine had to be able to simulate human touch, perform up to 3 tests at a time, be shippable, and cost less than $3,000. The Cycle Tester was a 3 month build - I was responsible for the mechanical design, electrical engineering, and programming.
After interviewing engineers to figure out what they needed, I discovered their biggest pain point in existing machinery is knowing precisely when a mechanism fails.
With these specifications and insight, I designed the Cycle Tester to drive electric linear actuators coupled with a force feedback system, that actuates push/pull cables.
The force feedback system highlights when there is a significant change in the performance of a mechanism. Whether it be a catastrophic failure, or gradual plastic deformation, these changes are all recorded using a data acquisition module (DAQami).
12V DC fans were added to keep the motors from overheating in order to prolong their life.
The frame was made from 80/20, and the front/back panels are custom designs. Switches were needed to manually set the limits of the actuator. I designed the fixtures and had them milled in China.
Force feedback couplings
Data from Force Sensitive Resistors highlight significant changes in performance of a mechanism or member.
After testing the power consumption of each motor under max load, I was able to spec the power supply. All 5 motors were driven using H-bridge chips connected to logic on an Arduino Mega.
I designed the PCB in Altium Designer. Large power traces were needed due to high power consumption of each motor.
Solidworks Assembly - Custom designed fixtures. Switches, buttons, motors, fans, and 80/20 models were imported.
In some use cases, screwdrivers are not as ergonomic or efficient as they could be.
Hext is a redesign of the screwdriver, giving the user more flexibility and power than traditional tools.
Hext has two sinuous gears which allow for a transfer of torque at various angles. This gives the user the freedom of using the screwdriver how they feel comfortable.
Surfacing done in NX. Renders in Keyshot.
This screwdriver is more efficient because it allows the user to both twist and revolve the handle around the axis of the screw. This design offers unmatched ergonomics and more leverage than any other screwdriver.
Early prototype, proof of concept
The tool should accommodate the user’s body, not the other way around.
The silicon mold was created using 3D printed parts.
3D Sintered parts were over molded with urethane.
A little flash, but nothing some handwork can't fix
This 1 week project showcases my 3D modeling and surfacing skills. I sourced 2D drawings of the car from Google images, and built the body as one part using sweeps, lofts, and other surfacing techniques in Solidworks.
I made the rims and tires separately and combined all parts in a final assembly.
The renders were done in Keyshot.
Varicare is a wearable medical device designed to alleviate the discomfort associated with varicose veins.
When a vein becomes varicose, its valves stop working properly, allowing blood to flow in the wrong direction. This creates high pressure on the venous walls, resulting in crippling soreness.
Varicare offers relief by massaging the leg in a peristaltic manner, creating blood flow while supporting the veins.
Medical studies on pressure therapy for varicose veins have found 80mmg of pressure to be effective. To meet this specification, while keeping the size of the motors to a minimum, the tension in the band had to be specified. Modeling the leg as a pressure vessel, and using 80mmg of pressure, the hoop stress (tension in band) was specified.
Varicare was prototyped using an Arduino. This microprocessor drives linear actuators while monitoring the force sensitive resistors under each band, controlling how much pressure is exerted.
Early concept: Motor housing suspended by bands and compression stocking.
Early concept: Brace with internal bands and pressure sensors.
Early Concept: Piezoelectric material to contract upon electrical signal.
Tension in band is determined using specified pressure and given assumptions.
Required torque is determined from the tension in the band and screw parameters.
Calibrating the pressure sensors
Known weight and area = Known pressure
Serial Monitor
