Avi Steinberg

A Node.js web app using TypeScript that allows users to visualize and manipulate how hair simulation, the cosmetic hair tattooing technique, would look on the client's head. I utilized D3.js to create spray brush and drawing tool functionalities on the client's photo. The image can be edited and stored using HTML5 Canvas. This app is used as a marketing tool to show potential clients what their hair would look like after completing the hair simulation procedure.

A Django web app using Python for the back end, a PostgreSQL database, and JavaScript, HTML, and CSS for the front end. I developed this app as a project for the knowledge-sharing committee at CKM Analytix. It is connected to Microsoft Authenticator, so employees can log in with their email addresses and update their technical skills. This app aims to make it easier to perform analytics on the employees' skills and identify strengths and weaknesses.

Before this app existed, the employees filled in their skills on a questionnaire, and the company saved and analyzed the data in CSV files. With the app, employees can update their skills in their profiles; for example, change a skill from beginner to intermediate when they complete a project. The data is stored in a database that enables analysis. I also created charts that visualize employee skill distribution for categories and subcategories. The charts update dynamically as employees update their skills.

A Python package and accompanying back-end code to perform data manipulations on call scripts and a Django application to present call center data using real-time visualizations. Our data engineering pipeline was directly connected to data generated from an Amazon Connect call center instance.

While co-developing this from end to end with a team of three, I wrote AWS Lambda functions to read, manipulate, and clean all call and agent data created by Amazon Connect and then write them to a PostgreSQL database. We connected the database to the Django models so we could use the data to create views of what was happening. One Lambda function was also connected to the Amazon Connect live streaming, and we were able to show live-stream call transcripts as they occurred with Amazon Connect agents.

We employed various natural language processing models—such as topic modeling and sentiment analysis—to provide more insight into analytics about categories of calls and to see which calls went bad. I worked on this project from end-to-end. We built this project as a prototype to showcase a proof of concept before the DevOps team incorporated the idea into their enterprise app.

Co-developed a Node.js process mining library written in TypeScript with the functionality to filter, visualize, and identify inefficiencies in large, complex, business operations processes. We implemented the Fuzzy Mining algorithm and other filtering algorithms we had developed to take raw audit logs and ticket data from IT incidents and service requests and categorize and filter them to show the visualization of the big picture of what was happening in the process. We provided a work group-level and an agent-level process map with summary statistics on each node and edge.

The visualization is interactive and gives the user the ability to start with a zoomed-out view of the process and then zoom in on any specific aspect of interest. We utilized D3.js with Graphviz to create process maps and other visualizations. We added sensitivity tuning parameters to adjust the degree to which users want to filter the process map.

This library is currently used by T-K, CKM Analytix' enterprise software, to power its process mining visualizations.

A tool that used R to scrape all product reviews for a given item on Amazon.com and then used analytics to create automated reports in Excel with sales projections about that product based on the frequency of recent reviews. These projections were based on the assumption that a certain percentage of people write reviews and used that to infer how many people bought the product. This provided additional insights to how products were doing from a financial perspective.