Nov 2024 – Feb 2026 · 1 yr 3 mos · Gurugram, Haryana, India · On-site

• Revamped cart's entire offer section to unlock the visibility of higher discount vouchers and nudge user behaviour towards increasing the cart value so as to unlock the next visible voucher. This was a key project to increase NOV (Net order value)
• Auto application of eligible payment promo. Resulted in increase in overall OTR (order through rate). This also lead to cart open/refresh P99 latency reduction by 15%.
• Migrated multiple high throughput components from monolith to micro-service in order tracking page which receives peak traffic of 70K RPM. Migrated - Notifications 15 million per day, Live activities 1 million per day, order status Aerobar in app 2.5 million per day with zero downtime and proper reconciliation mechanisms in place.
• Developed feature to allow rescheduling of scheduled orders. Introduced android live activity
• with launch of android OS 16. Developed interactive notification where user can specify delivery
• instruction directly by interacting with order status notification.
• Revamped entire Menu Management System of Zomato's "Everyday homely meals" segment [now closed]. Migrated the data management from sheets to SQL. Provided the flexibility to customise menu at Pod (individual store) level

Apr 2024 – Oct 2024 · 6 mos

• Backend Engineer in Risk-as-a-Service team that offers Fraud Protection solutions for transactions received by PayPal or Braintree merchants.
• Worked as part of team to externalised existing Fraud Protection Tools to sellers in marketplaces which resulted in revenue growth.
• Enhanced Transaction Reconciliation Process
• Played pivotal role in supporting migration of Big Query pipelines from DataProc jobs to BQ
• Subscription jobs from our team. Reduced annual cost by $250,000 and eliminated 5 mins of latency for every batch run.

Aug 2022 – Mar 2024 · 1 yr 7 mos

• Developed Model Score Normalisation Feature. Multiple machine learning models calculate risk
• score for a transaction. Whenever a model is changed/upgraded for a merchant it becomes a point of confusion for them. This feature normalises and map different scores that represent same risk to single score.
• Built Live Testing feature as a part of team where merchants can split incoming transactions on different set of filters by specifying percentage threshold and time duration, and evaluate performance of filters in real-time.
• Enhanced performance of transaction reports by optimising and partitioning Big Query Tables
• which resulted in reduced latency in execution.

Jan 2022 – Jul 2022 · 6 mos

• Worked as a Full Stack intern and developed an internal application for CRUD Operations of Variables.
• Developed UI interface using React framework and created a Microservice in backend using Spring
• Boot to serve the UI.

Sep 2020 – Feb 2021 · 5 mos · India

• Managed various sub-departments like perception, path planning, state estimation etc. as head of software department.

Aug 2019 – Sep 2020 · 1 yr 1 mo · Greater Delhi Area

• Developed perception stack for self driving vehicle using YOLO for object detection.
• U-Net deep learning model as a denoising autoencoder for lane detection.
• Data for training was annotated manually and Parameters for training model were decided empirically.
• Sensors like odometer, wheel encoders and GPS were integrated with help of to improve localization and data correction.

Aug 2019 – Nov 2019 · 3 mos · Greater Delhi Area