Jun 2023 – Jan 2026 · 2 yrs 7 mos · Remote

• Owned and maintained the entire infrastructure for Outerbounds, the enterprise offering of Netflix’s AI/ML tool Metaflow and an early-stage startup, while also personally building multiple key features. Regularly interfaced with customers to support onboarding, address customization requests, debug and resolve issues, and provide expert guidance.
• Independently built multi-cluster Kubernetes peering feature across AWS, Azure, GCP, CoreWeave, and on-premises environments using an experimental open-source project as a foundation and implementing a custom controller in golang. This innovation allowed customers to leverage diverse compute resources (e.g., scarce GPUs) for AI/ML workloads, significantly enhancing Outerbounds' competitive edge and driving customer acquisition.
• Spearheaded multi-cloud support initiatives, re-architecting and implementing the Outerbounds stack for Google Cloud (GCP) and Azure Cloud. This expansion enabled the company to secure new customers previously restricted by AWS only cloud offering.
• Designed and implemented OIDC-based authentication and authorization (authn/authz) for external Identity Providers (IdPs), enabling users to trigger Metaflow workflows via CI/CD tools such as GitHub Actions and CircleCI. Added integrations for external systems like Snowflake, allowing Metaflow jobs to authenticate using Outerbounds identity securely.
• Instrumented Highly Optimized critical components for better performance, such as optimized dependency builds using containerized workflows and a custom image registry backed by Object Storage services. Enhanced Kubernetes worker node performance by instrumenting optimized image pull integration with containerd runtime.
• Implemented an integrations feature, including a secure secrets management system, enabling seamless connections to external platforms like OpenAI, Hugging Face, and other third-party systems with user-friendly workflows.

Nov 2021 – Jun 2023 · 1 yr 7 mos · Bellevue, Washington, United States

• Created a simplified layer of resource defintion on top of Pulumi golang using go templating and yaml which reduced the code base size by 80% and gave users an intuitive API to make changes.
• Developed, scaled and maintained a secrete management infrastructure, HashiCorp Vault, with 150+ clusters across 3 cloud providers using infrastructure as code with Pulumi golang.
• Played a key role in scaling and maintaining a web of 170+ kubernetes cluster across 3 cloud providers and configured networking, monitoring, alerting, etc for them.
• Created and configured secret management infrastructure for a FedRamp High compliant environment

Mar 2018 – Oct 2021 · 3 yrs 7 mos · McLean, Virginia, United States

• Built a brand new AWS Organization from scratch, configured enterprise class networking stack per Organization Unit and created a mechanism for new AWS accounts with networking stacks to be created on-demand
• Worked on hardening and securing infrastructure such as AMIs, Netowrks, etc for SOC II compliance using tools such as Nessus, Ansible and automated CI/CD pipelines
• Configured and optimized new infrastructure including the networking stack, IAM, EKS, RBAC, Network Policies, and CI/CD pipeline for a lift and shift of the product into Kubernetes.
• Built internal Kubernetes clusters with dev and prod environments and provided Namespace as a Service with a CI/CD pipeline for engineering teams to deploy, maintain and own their internal services.
• Served as a dedicated technical advisor to 3 teams that built and maintained infrastructure and tooling and also consulted and advised other teams on architectural decisions

Jun 2017 – Aug 2017 · 2 mos · San Ramon, CA

• Developed and published the very first two AWS apps on the Phantom Security platform that automate the containment process of a compromised EC2 instance and access keys
• Explored the machine learning capabilities of Splunk and identified possible use cases for the Predix Cyber Security Incident Response Team
• Wrote search queries using SPL to preprocess AWS Cloud Trail and Cloud Foundry data, built Machine Learning models using classifier, anomaly detectors and clustering algorithms, and created a dashboard displaying the results of the ML models and their accuracy for further investigations by analysts

Oct 2016 – May 2017 · 7 mos · University of Maryland, College Park

• Quantitative Comparison of Modern Data Center Network Architectures, which focuses on the quantitative trade-offs of a variety of data center network architectures.
• Support research on available network modeling and simulation open-source tools, and explore ways to apply them and extract useful information.
• Additionally, setting up and configuring a network modeling and simulation environment analyzing the behavior of the flow and packet level of topologies, and different topology trade-offs.

May 2016 – Aug 2016 · 3 mos · Arlington, VA

• Used paragraph to vector and word to vector algorithms and classifier neural networks to compute and compare sentiment analysis capability on 1.6 million tweets.
• Setup and used Elasticsearch, Logstash, and Kibana to assist the development of offline GDELT like entity and wrote a python script that uses Hypnos to enhance its event recognition capability from text files.

Oct 2015 – May 2016 · 7 mos · Suite 210B, 8561 Fenton Street Silver Spring, Maryland

• Drupal Development
• Patch Panel work
• Organizing events
• Group Leadership and Organization