Our team at AQR computed certain parameters for the mutual funds, these values need to be validated by checking whether they fall in a standard distribution. This was done manually using Excel, the data comes in via REST API, which was extracted using Postman and then populated into an Excel file. Implemented this in Python as these checks took too long and were prone to manual errors. When gathering requirements, figured out that this can be implemented as a framework that other teams can use as well. Wrote a framework as a REST service where the user can specify the check to be applied and how to fetch the data, the framework provided connectors to fetch data from various sources (S3, REST API, DB, files), it also provided basic checks (standard distribution, greater/less than), provided base check classes to be overridden for custom implementation. The framework would apply the checks, store intermediate and final results, and these can then be viewed by another REST call.

In JPMC, modelers use Excel for data modelling, but this is not scalable or even trackable. Wrote a platform/app that could be used as an alternative to Excel using Jupyter Lab for UI and the back end was implemented using Python (Pandas, NumPy, Xarray), providing REST APIs (Django) for users to create/share/delete projects. The Jupyter Lab would connect to a kernel that had our library already imported, it also had a tab to view the tabular data (easier for folks who have always used Excel). This library contained function with function names and signatures similar to Excel so that it becomes easy for modelers to use it. Developed one key feature that was pre-loading user data into the Python kernel and assigning it to variable names defined by the user.

The project involved a setup where clients can host a quiz for their employees and record the responses. The data is stored across different tables. Wrote a framework to process the results and generate an appropriate alert based on various conditions. Wrote optimum SQL to query the PostgreSQL DB to fetch only the required records and not all. Processed this data with Pandas and separated it into different sections based on conditions for generating alerts. These alerts are then sent to the employees via an in-app notification and SMS. Wrote a Django UI for controlling the process so that new alerts/conditions can be uploaded and for configuring clients and their notification channels.