Neil Vinod

Work History

Wurl, Inc.

01.2023 - 05.2023

Translated UX and business requirements into elegant code solutions that were responsive on both mobile and desktop devices.
Created visualizations in Sigma (Tableau alternative) using data from Snowflake to illustrate the effectiveness of specific ads
Worked with product managers to develop the Adpool Portal app consistent with the designs created on Figma. We used Typescript/React for the frontend and AWS Amplify for the backend. Customers using the portal could self-service their accounts to change account info, purchase ad slots, and view ad metrics. I developed most of the frontend components/pages as well as many of the GraphQL resolvers that communicated with the database and Snowflake.

Amazon

01.2021 - 12.2022

Cut time spent interpreting Camera Search Metrics data by 80% by building the automated CS Metrics pipeline (AWS CDK pipeline). It sampled customer sessions, sent sessions as Amazon mTurk tasks to annotators for labeling, processed results, and delivered final data to applied scientists. I also built a web application to visualize the data using React and a couple visualization libraries (tables, line charts, spider charts, filters).
Created a web application that provided a centralized store for metrics to applied scientists. Using the app, scientists could re-run previously deployed ML models using updated parameters and quickly get new predictions as either a csv file or stored in a SQL table. React frontend and API Gateway/Lambda/DynamoDB backend using AWS CDK.
Built a visualization application for dataset leads to analyze previously labeled data and identify biases. They could also filter annotations done by an individual annotator to grade and identify possible repeated errors made by an individual. Leveraged Python/Pandas in the AWS API Gateway Lambdas to fetch and process the data and vanilla Javascript/HTML/CSS and Flask to display the visuals.

Amazon

Palo Alto, CA

01.2019 - 12.2021

Developed web annotation tools (Javascript, HTML, CSS) used by annotators to label data, which was then ingested by machine learning models to improve performance. Most of the tools were web pages that used both HTML Canvas and a variety of input fields to get bounding box and category data of the target items. The results were saved in large csv files.
Created a library of Python (Pandas) scripts to process annotation data in annotation metrics pipelines. The libraries were used to interact with different services, including our proprietary face-blurring service and a service to hide personal text information such as shipping labels and credit cards, and store the results in a shared AWS s3 bucket. The libraries increased efficiency and nearly doubled the number of images we could process in a week.

TapTrust

San Francisco, CA

08.2018 - 12.2018

Created the homepage, login page, and MY Account page of the React Native mobile app for TapTrust, a blockchain payments company
Designed and updated layouts to meet usability and performance requirements.