Mar 2024 – Dec 2024 · 9 mos · Gurugram, Haryana, India · On-site

• At Trademo, I worked as a Senior Backend Engineer focused on backend product development and optimization. My role involved improving technical workflows and designing a product from scratch, taking it through architecture, development, and production deployment.
• 1. One major challenge I tackled was resolving an Out of Memory (OOM) issue in the email generation flow. Users couldn’t download more than 75,000 records at once due to high memory usage and slow processing, caused by large Elasticsearch queries and sheet generation tasks. To address this, I implemented batch processing and optimized Celery by setting max-tasks-per-child = 1, ensuring memory was freed after each task. This allowed efficient task handling on limited resources, with 2 CPU cores processing at most two tasks simultaneously.
• These optimizations not only improved system stability and enabled seamless large record processing but also reduced memory allocation from 4 GB to 2 GB, cutting infrastructure costs while enhancing user experience.
• 2. The Finscreen application is architected to enable comprehensive risk assessment, advanced screening, and efficient data processing, with a primary focus on identifying potential red flags in financial and trade datasets. Leveraging a microservices architecture, I utilized Python (Django) for API development, Elasticsearch for distributed search and analytics, Redis for caching and real-time data storage, and Memgraph for graph database operations. To ensure asynchronous processing and scalability, I wrote an async engine capable of processing 50+ functions in a single run using an event loop with async, await, and asyncio in Python. This, combined with RabbitMQ for message brokering and Celery workers for task orchestration, enabled seamless data ingestion, high-throughput processing, and rapid scaling to manage complex workflows efficiently.

Apr 2021 – Mar 2024 · 2 yrs 11 mos · Gurugram, Haryana, India

• At Delhivery, I worked on large-scale event-driven architectural systems in the logistics domain, handling millions of shipments and ensuring seamless, high-performance operations.
• 1. Billing and Invoice Management
• Designed and developed a Billing Service to automate invoice generation for B2B clients.
• Triggered by waybill pickups at the First Mile or deliveries at the Last Mile.
• Managed billing details, supported invoice overrides, and consolidated invoices during billing cycles.
• Processed over ₹70Cr+ worth of invoices monthly with seamless automation.
• 2. PDF Generation and Dispatch Services:
• Built a PDF Generation Service for invoices, LR copies, and Last Mile POD copies.
• Used Amazon S3 for document storage and shortened URLs for efficient access.
• Generated PDFs for over ₹70Cr+ worth of B2B invoices monthly.
• 3. Invoice Dispatch Service (B2B Clients):
• Developed an Invoice Dispatch Service to physically dispatch invoices, mimicking shipment flow from First Mile to Last Mile. Enabled seamless tracking and payment collection for clients.
• 4. Modern Architecture and System Migration:
• Architected systems using a modern tech stack: AWS Lambda, Kafka, Elasticsearch, DynamoDB, S3, Redis, Flask, EKS, ensuring scalability, reliability, and operational excellence.
• Led the migration of several services from serverless architecture to EKS (Elastic Kubernetes Service).
• Enabled dynamic pod scaling based on CPU usage to handle high-throughput workloads during peak times. Achieved cost efficiency and improved resource utilization for critical services.

Aug 2019 – Mar 2021 · 1 yr 7 mos · Bangalore

• At Danske IT I was a part of Big Data and Advanced Analytics Team.
• Data Pipeline: Designed and Created a Data Pipeline where we are transferring Data from Hadoop to Data Warehouse I have worked on complete designing of Data pipeline from developing the pipeline from scratch to deployment in production.
• Technology used: Python, Hive, Sqoop, Pipeline(Jenkins) , Git, Airflow (used for scheduling)
• Oozie Job Tracker: Designed and created an application that will detect the status of all Oozie jobs (workflow jobs) before and after the job completion on a daily basis(Similar like CRON Job). We can keep track of jobs and process them again if needed, this application saving time as before we manually check for all jobs that are ingesting data in Hadoop.
• Technology used: Python, Oozie, Git, Jenkins(Pipeline) , Airflow .