A full account of where I've worked, what I built, and what I learned. I care more about the problems I solved than the titles I held.
Current
Jan 2025 - Present
Hybrid - Bengaluru
Okta
Software Engineer
Building AI platforms and developer infrastructure at Okta. Most of my time goes into agentic systems that help engineers investigate production issues, automate ops workflows, and move faster on cloud-native work across 18+ AWS production environments.
Key Contributions
Built Infra AI, an internal AI platform integrating infrastructure systems, operational knowledge, and production tooling for contextual infrastructure search across 22+ AWS production environments.
Developed production AI agents with Amazon AgentCore and MCP servers for Spinnaker, AWS, Kubernetes, Terraform, Grafana, Datadog, and PagerDuty.
Built an AI-driven Terraform plan intelligence service on AWS Bedrock using semantic chunking and hierarchical summarization for deployment summaries and risk assessments at 1,000+ release trains/day.
Built The Guardian, an internal agentic AIOps assistant integrating Slack, Confluence, Amazon Kendra, and platform services for on-call incident assistance.
Scaled Infra AI from 50 to 1,000+ users with conversational memory, session management, and UX improvements.
Migrated AI agent infrastructure from AWS Lambda + ALB to Amazon AgentCore Runtime and Agent Gateway.
Reduced infrastructure drift by 70% across 18+ AWS production cells with a distributed Go-based cloud-agnostic drift detection service.
Improved deployment throughput 4x (8 hrs to 2 hrs) with a Spinnaker orchestration framework coordinating thousands of concurrent pipelines.
Owned and maintained 5+ business-critical platform services for deployment orchestration, infrastructure automation, and SRE workflows across 18+ production environments.
Built payment and fulfillment backends for a fashion commerce startup incubated at VIT. Focused on checkout reliability, order lifecycle APIs, and a schema that could grow with catalog and logistics complexity.
Key Contributions
Integrated Razorpay end-to-end for checkout and settlement flows, enabling live payment processing for the storefront.
Integrated Shiprocket APIs for order creation, tracking, and returns, closing the fulfillment loop from purchase to delivery.
Designed and implemented the core database schema for catalog, orders, and logistics data to keep reads and writes efficient as volume grew.
Technologies
Node.jsRazorpayShiprocketSQLREST APIs
Internship
Oct 2023 - Mar 2024
Remote
Samsung R&D India
Software Development Engineer Intern
Worked on image processing and ML pipelines for astrophotography. Focused on quality improvements and making the processing path faster and leaner.
Key Contributions
Improved astrophotography image quality by 35% using CLAHE and adaptive ML-based tuning while reducing pipeline memory usage by 40%.
Technologies
PythonOpenCVMLCLAHEC++
Internship
Aug 2023 - Oct 2023
Bengaluru
Carl Zeiss India
Summer Intern
Built data pipelines and reporting automation for multi-region operational datasets.
Key Contributions
Developed Python ETL pipelines processing 10k+ multi-region records and automated KPI dashboard generation, reducing manual reporting effort by 30%.
Technologies
PythonETLSQLDashboards
Internship
Jun 2023 - Jul 2023
IISc, Bengaluru
AutoSky Aerospace
ML Intern
Built computer-vision pipelines for Foreign Object Debris detection on aerospace surfaces, from classical OpenCV detection through CNN classification and edge deployment on Jetson Nano.
Key Contributions
Engineered the FOD detection pipeline from scratch, starting with OpenCV algorithms for the initial detection stage.
Integrated OpenCV detections into SqueezeNet and DenseNet classifiers, reaching 94% accuracy while improving inference speed by 50%.
Implemented YOLOv5 for FOD detection and classification, cutting Jetson Nano runtime by 30% while holding 91% accuracy.
Technologies
PythonOpenCVYOLOv5SqueezeNetDenseNetJetson Nano
University Team
Nov 2021 - Feb 2023
VIT Vellore Research Team
Team AutoZ
ML Engineer
University technical team building an autonomous rover for IGVC. Worked on perception for lane detection and obstacle segmentation in a 12+ person squad, with a heavy focus on coordination across software, hardware, and controls.
Key Contributions
Contributed perception systems for an autonomous rover that placed 4th at IGVC, USA.
Built lane detection and custom obstacle segmentation for real-time avoidance on the competition course.
Worked on ROS-based simulation of autonomous capabilities in Gazebo to validate perception and navigation before hardware runs.
Shipped product surface and ops automation for a carpet-based NFT marketplace, from a responsive landing experience to bots that handled collection prep and minting.
Key Contributions
Built a responsive NFT landing page for a carpet-based collection, improving mobile load and mint-path clarity for marketplace traffic.
Developed Python automation bots for image scraping and OpenSea minting, reducing manual collection ops from multi-hour sessions to minutes.