Work Experience

Software Engineer @ Databricks

Performance Engineering (PerfEng) Team (Oct 2024 — Present)

• Working as a Software Engineer on the Performance Engineering team at Databricks

Senior MTS Software Engineer @ Salesforce

Salesforce Database Team, San Francisco, CA, USA (Jan 2024 — Oct 2024)

• Architected and developed real-time observability dashboards using Grafana and Superset, leveraging database engine telemetry to provide critical performance insights for a large-scale Postgres deployment
• Engineered and executed comprehensive database workload benchmarks to evaluate and integrate new database engine features, directly influencing product performance and scalability
• Designed data models and built ETL pipelines using Python and Airflow to process and analyze performance data, identifying and resolving critical system bottlenecks
• Technologies: Postgres, Java, Python, SQL, JavaScript, Trino, Airflow, Superset, Grafana

Graduate Student Researcher @ UC San Diego

UC San Diego, La Jolla (May 2023 — Sep 2024)

• Developed an error detection tool for spreadsheets under the guidance of Prof. Michael Coblenz
• Technologies: TypeScript, React, Python

Web Developer @ UC San Diego (Part-time)

UC San Diego, La Jolla (Oct 2022 — Apr 2023)

• Developed web applications for music visualization
  Check it out
• Technologies: React.js, JavaScript, HTML, CSS, TypeScript

Software Engineer @ Salesforce

Hyderabad, India (Jun 2021 — Sep 2022)

• Designed and developed the extensible Salesforce Scheduler record-sharing backend service, fully automating the sharing process. Wrote the module to be generic enough to be reusable by other teams, since record-sharing is a core part of the Salesforce platform.
• Developed the backend and frontend for recurring shifts, which improved the shift creation process by 300%. Referred to RFC 5545 (iCalendar) for implementing the recurrence pattern. Made data model changes for incorporating creation of recurring shifts and storing the recurrence pattern. Developed the user interface for creating daily, weekly, and monthly recurring shifts.
• Developed the troubleshooting tool for Scheduler, aimed at quickly resolving customer investigations. The tool is useful for developers during data setup, customer-centric engineering, and for the support team to debug customer investigations.
• Worked closely with the PM on the design and documentation of a prescriptive Scheduler data setup for Automotive Cloud use cases
  Check it out
• Improved code coverage (line and branch) of five modules, averaging five files per module with over 100 lines each, to over 90%.
• Technologies: Java, JavaScript, Lightning Web Components, Databases, Salesforce Platform
One of my most challenging and interesting projects...

One of the projects I worked on during my tenure at Salesforce was the recurring shifts feature on the Scheduler product. I owned the entire feature and designed both the frontend and backend. The feature was quite complex and involved brainstorming to tackle various practical scenarios. It required data model changes, and the implementation was also challenging. It made the shift creation process 300 times more efficient. I wrote modular code with best coding practices and tested the functionality thoroughly by writing multiple unit and functional tests. There were some discoveries that I made during the last half of the release related to timezones. I was able to present a solution that was approved by the seniors and PMs and implement it. I learned both technical aspects like data model changes, best coding practices, and thorough testing, as well as non-technical aspects like tackling and handling last-minute discoveries calmly.

Software Engineer Intern @ Salesforce

Remote, India (May 2020 — Jul 2020)

• Was part of the Consumer Goods Cloud (CGC), specifically the retail execution team
• The objective of the project was to create a dataset generation utility capable of taking an initial set of sample product images and a few shelf images to serve as backgrounds, then producing a robust, diverse dataset by placing the products on the shelves, applying various augmentations, and generating a dataset containing the images and a CSV file with the annotations of the products in each image.
• Creating a dataset for an object detection model is an expensive task as it involves manual effort to identify the products on each shelf, determine their position, annotate them, and prepare the training dataset.
• Developed a dataset generation utility to produce a large, robust, diverse dataset from a limited sample of product images
• The utility reduced the manual effort involved in annotation of various objects in images and made the process 630 times more efficient
• Product detection accuracy of the Einstein Vision model after training with the dataset generated by the utility was observed to be 98.7%
• Wrote modular code using OOP principles in Python; incorporated memory profiling, time profiling, logging, and configuration management
• Technologies: Python, OpenCV, ImgAug, Augmentor

Teaching Assistant @ IIT Hyderabad

IIT Hyderabad, India

• Teaching Assistant for Data Structures-I, Algorithms, Introduction to AI courses
• Responsible for conducting labs, guiding students in their labs, evaluating quizzes, final exams, coding assignments