Bounce Rate Optimization with Markov Traffic Models
- We address the following problem: In a network if items and tags, how does one choose tags for a new item to maximize probability of a user reaching that item?
- After coming up with, refining, and culling a few model modifications, we arrived at a variant with a different weight context and objective for which the problem
is still NP-hard. This model is also more realistic, since it is incentivized to maximize overall engagement, not just with the new item.
- Read more about our findings in the presentation and
TextWifi: Accessible Wifi for All
- Inspired by the sharing economy and the on-the-go lifestyle of college students in NYC, TextWifi taps into previously untapped sources of wifi, such as public
networks (e.g. LinkNYC, Downtown Brooklyn, MTA), as well as private guest networks (e.g. Starbucks) to provide a unique platform for connecting
community and connecting to Internet
- Built using Python/Flask, with Amazon AWS DynamoDB + ElasticSearch and the Twilio API.
- Check out the presentation and
- Lyric Machine is a Chrome extension that makes it easy for you to sing along to your music.
Simply click on the icon while listening to your YouTube song and get the lyrics (and
a bonus gif) instantly!
- Built with the Musixmatch and GIPHY APIs.
- Created at PennApps XVI 2017. Check out the DevPost and get it on the
Chrome Web Store.
- PetPlay matches people who want to spend time with pets, for anything from hiking to watching
TV on the couch. PetPlay solves the immediate need of pets at shelters that need to be
taken care of, and increases fostering and adoption rates in the long run.
- Consists of a mobile user-facing app and a web shelter-facing app. Created using Objective-C,
Node.js and Firebase interacting with the PetFinder API, Pitney Bowes API, Button API,
and Sendgrid API.
- Winner of Most Commercially Viable Solution using Pitney Bowes APIs at Treehacks 2017.
- Check out the DevPost and demo the shelter-facing
side on Heroku at playwithpets.herokuapp.com.
- Using a comprehensive database of drug effects (Offsides) and a database of drug-drug interaction
side effects (Twosides), this web app visualizes drug relationships and effects in a
force-directed graph and adjacency matrix. It is a useful tool for researchers and clinicians
to identify key interactions between drugs.
- Created using Python/Flask interacting with data provided by Professor Nicholas Tatonetti and his lab at Columbia University. The app also uses D3.js for visualizations.
- Check it out on Github and demo it on Heroku at binf-visualizations.herokuapp.com.
- This web app allows you to search an NBA player and see how his activity on social media
affects his performance as a teammate on the court, measured by five different statstics.
We present the data compared to the rest of the team.
- Created using Python/Flask interacting with the Twitter API, with Chart.js for visualizations
and data provided by the NBA for the inaugural NBA Hackathon in NYC.
- Check it out on Github and demo
it on Heroku at nbahackathon.herokuapp.com.
- Leaplock allows users to use 3D hand gestures to unlock a safe or any kind of lock. Users
can register hand gestures as a "password" gesture.
- Built using the LeapMotion Controller to track hand gestures, Node.js with Socket.IO for
the web interface, and Beaker Notebook for additional pattern visualizations. Also used
Google Firebase to store data.
- Honorable Mention Winner for Best Use of FOSS in a Drone Cargo Management System App by Local
Motors Labs at HackMIT 2016.
- Check it out on Github and DevPost.
- BfB is a Tinder-like app that aims to solve food waste/hunger problems by letting people
easily broadcast that their food is about to be trashed if not eaten soon.
- Winner of best UI/UX and most practical at Facebook Hack for a Cause Seattle 2016.
- Check it out on Github and demo it on Heroku at
- Created an app that allows students at Cal Poly Pomona to easily find out prerequisites for
courses in the Mechanical Engineering Department.
- Tools used: Node.js, Express, MongoDB, Bootstrap.
- Deployed with MongoLab on Heroku at cpp-courses.herokuapp.com and available on Github.
- Tracking disease prevalence and providing solutions using socioeconomic and healthcare data.
- Wrote several Python scripts to scrape a number of socioeconomic and disease related information
from online sources such as Wolfram Alpha and Twitter.
- Goals: understand how diseases travel among populations and provide society with warning
of incoming epidemics, help social infrastructures such as hospitals select socially
optimal expansion points.
- Tools used: Node.js, Twitter API, Wolfram Alpha API, Leaflet.js, D3.js, Materialize.
- Available live here and on Github. Made during HackHarvard 2015.
- The simple way to meet up with friends.
callbacks, and implement Materialize, a CSS framework.
- Available live at gocentr.me and
on Github. Made during YHack
- Conducted collaborative research work and web application development at Matsuo Weblab (GROW
Lab) at the University of Tokyo under Prof. Kotaro Nakayama.
- Worked on Twitter and Wikipedia data analysis as well as text mining using the Google Custom
- Implemented new features such as Authenticated Social login, "Like" feature, and Google Trends
analysis for each title in a database of over 60,000 anime, game, and manga titles.
- Created a novel recommendations engine using collaborative filtering algorithms, optimizing
performance and accuracy.
- It's available online at asiatrendmap.jp.
- I spent eight weeks in Hong Kong as a product development intern working on a data visualization
project, taking over 360,000 database projects and displaying them as points on an interactive
- You can filter the projects by commodity, country, and number of projects.
- Created using Python, Flask, D3.js, and Leaflet.js.
- It's available online at analytics.thecloudminer.com.
- A website where friends, acquaintances, and strangers can schedule appointments with me.
- Created using the Google Calendar API, Python, and Flask.
- It's available online at bobawithjames.herokuapp.com and on Github.
- It's available online at mangaquiz.herokuapp.com and on Github.
- A simple exercise to get comfortable with Python and Flask GET requests, HTML/CSS, Jinja
templating, Foundation, and Heroku.
- It's available online at ronswanson.herokuapp.com
and on Github.
- Prototyped Savi, your personal coupon manager.
- For people who are looking to save money and don’t have the time to regularly coupon or search
for coupons in the clutter of their inbox, Savi is a coupon management system for iOS
with location-based coupon discovery that allows users to organize coupons in a clean
and easy-to-use mobile interface, and discover deals when they’re out and about.
- Features include saving coupons in a Wallet, and searching for coupons through Discover.
- Landing page with promotional video and social media links are available at savi.strikingly.com.
- Worked in a five-person team over 8 weeks to undertake comprehensive research and analyze
objective and subjective data to come up with strategies for Missing People, our charity
- Developed possible IT solutions, including prototyping a microsite as a separate zone for
youths, with key features such as an interactive map, chat room, FAQ, and forum. Emphasized
confidentiality and support.
- Included very specific estimates on the cost implications of our proposed solutions and avenues
that they can consider for funding.
- Made sure that Missing People's efforts to improve engagement with children and young people
and their overall digital technology strategy remains up-to-date with current times and
is robust towards a wider range of issues.
- Set out a roadmap for the future IT development for Missing People, including a mobile application
for iOS and Android.
- Learned industry consultancy practices from executives at TCS and interacting with representatives
from Missing People.
- Runner up for the Tata Consultancy Services (TCS) Tech Challenge 2015