Abhi is a junior at the University of Pennsylvania double majoring in Biology and Computer Science. He is passionate about creating social impact in the community using programming and technology (through his work with Hack4Impact). Additionally, he does research on Traumatic Brain Injury and is a developer and researcher at mHealth in the Hospital of the University of Pennsylvania. He also loves to teach through his course (CIS197: JavaScript) and his Medium.com article series. Outside of academia, Abhi is the president of the Penn Publications Cooperative (overseeing an overall circulation of 20,000 publications at Penn) and is a member of the oldest continually-existing literary society in the United States (the Philomathean Society).
Lecturer for course on Javascript. Responsible for lecturing class on weekly basis, creating homework assignments, and managing TAs.
Writing book on introduction to JavaScript covering everything from fundamentals to how to create full stack applications using React, Redux, GraphQL, Node, and Mongo.
Teaching assistant for course for 3 semesters prior
Developing portable mobile applications for computational diagnosis of Traumatic Brain Injury.
Built algorithm to localize pupils and calculate distance between eyes in response to visual stimuli using OpenCV + gradient ascent methods
Created Convolutional Neural Network to classify 7 different emotion in real time using 1700+ images from public facial expression databases (92% accurate)
Creating clinical study and developing React Native application for Apple ResearchKit apps
Researching correlation between use of web and mobile applications in transplant and post-transplantation survival rates
Built entire web application from scratch (with flask) to track patients' exercise, medication, and nutrition activity
Used AWS S3 to store research study videos & data. Also created configuarble system to schedule CRON jobs to notify patients about upcoming todos.
Integration with REDCap for HIPAA compliance
Built interface over SMS for patients without access to desktop/mobile computers using Twilio API
Hack4Impact is a student led 501(c)3 organization on UPenn's Campus helping create software to help local and national non-profits accomplish their missions.
Filed for 501(c)3 status for the club & incorporated organization
Managing group of 30+ students working with 5 nonprofits/semester
Spearheaded chapter program to nationalize Hack4Impact (as of Jan 2018 there are 7 chapters across the nation
500+ volunteer hours during 2017. Technical Lead (Spring 2016-Spring 2017). Product Manager (Fall 2017). Projects Manager & Executive Board member (2017). Co-Director of organization (Jan 2018-present)
Project for organization Madaktari Africa (on behalf of hack4impact) to create volunteer management and matching system
Worked with international client to create and launch app
Product Manager of team of 6 individuals, responsible for planning & setting project scope over 6 months
Implemented dynamic form creation, generation, and response management for administrators
API that uses the im2txt model to take images and output neural network generated descriptions
Reduced size of im2txt model and dependencies to under 512 MB to create Lambda API (previously 1.2 GB in size)
Reduced runtime of im2txt model to < 10 seconds (from 30 seconds)
Created corresponding chrome extension to be added as a browser tool for visually impaired individuals to use (abhinavsuri.com/aat)
A website to track tweets globally relating to the immigration ban
Using twitter streaming API and D3 JS created globe visualization of live streaming tweets under certain hashtags
Front page product hunt
Allow users to launch their own instance of Maps4All (a flask application with a backing DB and redis queue) and their own URL using Heroku OAuth and API
Technical Lead on Team of 6
Implemented initial Docker solution, then reworked to take advantage of Heroku's Dyno Launch API via Heroku OAuth
Prototyped dynamic subdomain URL generation + registration with namecheap
A tool for visualizing your complete, consolidated, collected Google Location History.
300+ github stars (as of Jan 2018)
Contribution allowed user to stream large location history file and display data on heatmap accordingly
Increased maximum filesize the application could process from 180 MB to 2.2GB using Oboe.js
Processing millions of Google Location Data points into court admissible legal documents...all on the frontend
This app processes millions of Google Location data points into a CSV Files based on 'bounding boxes' (i.e. ranges of time within regions of interest). The project is intended for use in worker dispute cases.
Technical Lead on Project Managing team of 6
Used Oboe.JS to allow large file upload and visualized data points by altering leaflet.js library to cluster data points map
A simple Flask boilerplate app with SQLAlchemy, Redis, User Authentication, and more.
1300+ github stars (as of Jan 2018)
Top contributor, maintainer, and creator of all documentation
Reverse engineered YouTube Caption API to get captions for videos and used Jaccard String Similarity to do fuzzy matching on user requested captions.
Grand Prize Video MLHPrime Hackathon
Front page product Hunt
Created Augmented Reality Education platform with Google Cardboard. Used marker based SLAM to calculate position of multiple viewers of marker and then overlayed 3D objects onto marker with correct orientation for all users.
Top 10 & Grand Prize AR/VR PennApps XIII
Created online market place that manage trasactions for over 10,000 different items
Technical Lead of Project Spring of 2016, managed team of 6 as a freshman
Created machine learning model for automated basal cell carcinoma assessment and diagnosis through Random Forest based ensemble training Methods with Computer Vision Processing.
Grand Prize Medicine & Health PennApps XII
Also minors in Chemistry and Classical Studies. Bachelor of Arts Degree from the College of Arts and Sciences.
Released on FreeCodeCamp (Top Medium tech publication) 28k views (as of Jan 2018)
Presented at RSNA 2017 (Cum Laude Award Winner: Top 65 of 1812 posters)
Released on Penn Engineering Blog (and newletter) 1k views and featured on homepage of penn computer science dept (as of Jan 2018)
Released on FreeCodeCamp and HackerNoon 15k views (as of Jan 2018)
Released on Towards Data Science 5k views (as of Jan 2018)
Released on FreeCodeCamp (top Medium tech publication) 10k views (as of Jan 2018)
Released on FreeCodeCamp (top Medium tech publication) 32k views (as of Jan 2018)
Scientific and Educational Poster presented at the 2016 Annual Meeting, Society of Abdominal Radiology
'The effects of perturbed cerebral blood flow and cerebrovascular reactivity on structural MRI and behavioral readouts in mild traumatic brain injury'
Scientific and Educational Poster presented at the 2014 Annual Meeting, Radiological Society of North America