Aug 2025 – Present · 9 mos

• After having spent a year in MLOps, decided to go back to the domain I am passionate about: Data Observability. The charter of the team includes: Data Lineage & Data System Health Monitoring.

Mar 2024 – Aug 2025 · 1 yr 5 mos

• MLOps (Machine Learning Operations): While machine learning is not new to LinkedIn, we are on an exciting journey to dramatically improve the productivity of ML engineers. My focus is on two key areas:
• 1. Enhancing Metadata Collection - Ensuring that LinkedIn continues to comply with various regulations by improving how we collect and manage metadata.
• 2. Automation of Training Pipelines and Models - Implementing CI/CD for training pipelines and models to reduce human toil, increase efficiency, and accelerate the development process.
• By driving these improvements, we aim to empower our ML teams to focus more on innovation and less on manual tasks.

Oct 2021 – Mar 2024 · 2 yrs 5 mos

• Team: Tracking at LinkedIn
• When I transitioned from Mobile Infra to the Tracking team, I continued owning the Tracking SDKs while expanding my scope to include the Data Processing of Tracking Data. The responsibilities of my team included:
• 1. Accurate and Consistent Data Collection – Ensuring timely, accurate, and consistent data collection of member engagements across all LinkedIn mobile and web apps. This was achieved by building native SDKs (Android, iOS, Web/Ember, and React) that automatically collect data, drastically reducing human error.
• 2. Data Processing – Processing the collected data to: a) Empower internal teams with the right metrics to make informed product decisions; b) Ingest the processed data into AI/ML workflows for quick decision-making to enhance the member experience.
• The Data Processing platform was built using a mix of technologies such as Spark (for offline processing), Samza/Flink (for nearline processing), Couchbase (for intermediate storage), and Airflow (for orchestration).
• I worked closely with the Data Infrastructure team, providing requirements, ensuring timely delivery of necessary functionalities, and continuously providing feedback to improve the customer experience of our platform.
• I also led a company-wide initiative to migrate internal customers to the new Tracking stack (SDK + data processing). Under my leadership, we successfully migrated 25% of the traffic from the old stack to the new one before transitioning to MLOps.

Oct 2018 – Sep 2021 · 2 yrs 11 mos

• Team: Mobile Infrastructure
• Vision: Provide a delightful native experience to LinkedIn users on all mobile platforms.
• Mission: Provide foundational components and systems to all mobile engineers at LinkedIn to help them develop faster, release seamlessly, monitor in real-time and continuously iterate to improve the product.
• As part of leading this team, I scaled the team from 5-member team in 2018 to 14 member team in 2021 and we delivered on multiple key infrastructure projects such as GraphQL, Intelligent & automated Impression Tracking, Monitoring Insights for Mobile Apps to name a few.

Mar 2017 – Sep 2018 · 1 yr 6 mos

• Team: Mobile Infrastructure
• Project: Client side experimentation library on iOS
• At LinkedIn, we continuously perform experiments to better understand user behavior and improve member experience. As part of this project, I re-wrote the client side experimentation library on iOS, which makes it easy for our developers to run experiments.

Jul 2016 – Feb 2017 · 7 mos

• Team: Service Infrastructure
• Project: Parseq Lambda Names (https://github.com/linkedin/parseq/tree/master/contrib/parseq-lambda-names)
• Parseq (https://github.com/linkedin/parseq/) is a framework that makes it easier to write asynchronous code in Java by introducing a notion of 'Task'. I worked on a project to provide more meaningful default descriptions for Parseq tasks. Using ASM, this project tried to locate where lambda expression is defined in source code and also infer some details about its execution like function call within lambda expression with number of arguments.

Oct 2015 – Jun 2016 · 8 mos

• Lead an initiative on providing a consistent way for members to report inappropriate content on LinkedIn. Involved building libraries for multiple platforms like iOS, Android, mobile web and desktop as part of frontend and unified backend which decides to send content for review depending on severity of reports.

May 2014 – Sep 2015 · 1 yr 4 mos

• Working with Content Filtering team. Major projects:
• 1. Integrated the spam processing platform with comments on site.
• 2. Designed and developed a unique process to continously develop and improve machine based models for detecting spam. Applied a patent for the same (currently in pending status). Involved developing a platform for sampling and labeling content, storing labeled content and calculating precision/recall of machine models.