Keya Shah

Qyu.ai is a startup that is building a causal inference platform aimed at estimating heterogeneous treatment effects in observational data. My responsibility is to work on the backend of the platform: -Set up and connected Firebase with Python Flask to save user's preference and selected variables -Connected Flask backend with React frontend using GET and POST request for data transfer -Worked on storage allocation and inserting files on GCP

-Helping Hands is a non-profit organization that helps deliver food through volunteers during the pandemic. -I worked on improving the user interface and the functionality of the console that manages the deliveries such as enabling the users to edit their delivery after its creation and updating the backend code to reduce wait times between deliveries. -Some of the languages and platforms I used are React and Firebase.

The Indian Student Association at UC Davis is an undergraduate student-run organization that brings the UC Davis community closer through holding social events for cultural awareness, fundraising for the underprivileged community, and working in conjunction with other South Asian Organizations. Every quarter we put on fun events for the UC Davis community such as Garba, Winter Formal, and Culture Show. My responsibilities as the Director of Tech and Design are to -Work together with other board members to organize events for more than 50 people -Lead and manage committees to develop programs to promote inclusivity and connectivity -Work with marketing directors to create posters and flyers to promote events -Teach and guide five interns in the roles and responsibilities of a director

At HackDavis 2021, my team and I created a web application that uses eye-tracking technology to give insight into a user's focus on a computer screen. In the Train tab, click on the nine points displayed on the page five times until their color turns yellow to calibrate the eyes. Test the focus In the Implementation tab with any youtube video. As soon as the video starts, a heat-map will be generated based on where the user is looking on the screen. The video runs for one minute, at the end of which the website will display an alert box with the percentage of the time the user was looking at the video -Used JS library Webgazer.js and FaceMesh to implement eye-tracking -Integrated JS, HTML and CSS to create a web interface to show proof of concept

For UPenn Hacks 2020, my team and I built a web-based Machine Learning project called Accessible Smart Linguist that detects American Sign Language using the live video feed from the camera and implements features like autocorrect and autocomplete to reduce the communication time between two people. Furthermore, the Learn feature of our application enables users to explore and improve their sign language skills in a fun and engaging way. -Self-created an image dataset for the 26 letters of the alphabet using OpenCV and python scripts -Trained a Convolutional Neural Network using TensorFlow and Keras on the dataset for 50 epochs achieving about 98% accuracy -Integrated the model into our website through the use of Python and Flask -The website takes in a live video stream from the camera and the user can sign in the designated space -The last word is autocorrected and autocompleted every time the user enters a space. The user can also manually click the autocomplete button -The user can start or pause anywhere as well as start with a clean palette -The Autocorrect/Autocomplete features are implemented through python libraries using an NLP model that trains as the user types -The front-end of the application is developed through the use of HTML and Flask templating

This hackathon was sponsored by HCL and Microsoft companies to encourage the development of applications that could help in the fight against coronavirus. My team and I created a mobile app called COVID Trace that allows the users to monitor their health, know their risk, and report symptoms. -Developed the app through Flutter packages and Dart programming language -Built the home screen with multiple resource links and an API that updates the global coronavirus count daily -Linked the self-reporting page to the home screen where users can report their test results or contact and made sure to validate the answers -Integrated the calendar page that retrieved the corresponding data from the answers and displayed a note each day for the next 14 days reminding the user to self-isolate and report any symptoms -The app includes prevention tips as well as reliable sources such as CDC and WHO -All the data is anonymous and stored on the user's phone to avoid any breach in privacy

At Hack Davis 2020, my team and I created a mobile application called VeggieBuddie. Through VeggieBuddie, you can tell whether a food item is vegetarian, vegan or non-vegetarian by taking a picture of the ingredients. Additionally, the app checks for eight major allergens that you can customize to your own needs in your profile page. -Built the app in Android Studio using Flutter and Dart programming -Used google user authentication so that the user can log into their Google account to save any dietary preferences, specifically allergens. -Implemented Google's API for Optical Character Recognition to extract the text out of the image and then compared it to the locally stored list of ingredients to determine the dietary restriction. -Saved the allergens list in the profile page in the google cloud realtime database, Firebase. -Offline access by logging into the guest account

Volunteered as Software Developer at Talix and performed the following tasks. -Worked on de-identifying documents to redact Personal Health Information (PHI) to comply with HIPAA -Tested CodingInSight application to identify performance issues -Decomposed complex SQL statements to easily understand Entity Relations -Used SQL Developer to build, populate and join tables through foreign keys for Testing -Queried tables with join statements for specific conditions frequently found in Medical Charts -Created dynamic SQL statements to prevent redundancy and improve performance time -Used Java to query the database and perform CRUD operations -Created CRUD equivalent REST API’s utilizing JAVA SpringBoot framework -Built a Front End application using JSP’s that uses the above REST API’s to Query, Edit and Delete records