Headline

Platform Engineer @ dunnhumby | Kubernetes (GKE, AKS) · Istio · Terraform · Go | Multi-cloud: GCP, Azure, AWS | Service mesh, DR, Workload Identity Federation · Bengaluru

About

I build and run the platform other engineers ship on. Five years across GCP, Azure and AWS — GKE and AKS, Istio, Terraform, Go tooling, CI/CD — owning provisioning, delivery and failover end to end. Currently on dunnhumby's 10-person core platform team. Some of the work: → dhctl, our team's Go CLI. Before it, every engineer wrote their own code to reach a cluster, raise a PAM request or discover a cloud resource. My idea, I founded it, all 10 engineers use it. I later led the epic that moved it off shelled-out exec calls onto the native Azure and GCP SDKs. → The Azure disaster-recovery failover edge for a 5-region multi-tenant platform: the path that carries traffic when Cloudflare is unavailable, cutover driven by a flag in code, designed for 5-minute recovery. I ruled out Azure Front Door after TLS-termination mismatches and asymmetric user-defined routes broke standby health probes, and shipped on Application Gateway. → A Terraform self-service platform on AWS at Thoughts2Binary: five lines of config provision a full stack — 3-AZ VPC with per-AZ NAT, ALB with Auto Scaling, ECR, KMS-encrypted S3, least-privilege IAM, CodePipeline/CodeBuild. Ten for_each modules, so a new service duplicates no code. Onboarding went from three days to under 30 minutes. First time I'd touched Terraform; I broke the platform repeatedly getting there. → MongoDB in production there: a 3-node replica set on EC2, one node per AZ, replacing a single-AZ failure mode with manual recovery. No restore path existed, so I built backup and restore from scratch and verified it the only way that counts — deleted the database, restored in 10 minutes. Outside work I'm building Transit (github.com/pratyushsinha05/transit), a Go geo-telemetry ingestion and live-broadcast engine. H3 resolution-9 cells as the spatial index, so geofencing is a cell-equality check; TimescaleDB and PostGIS for history, Redis for hot per-device state, WebSocket broadcast tested at 1,000 concurrent clients. I also write about platform engineering and the AI-agent harness — the deterministic systems around a non-deterministic core. Same discipline, one layer up. Looking for platform, infrastructure or SRE roles on product engineering teams in Bengaluru or Gurugram. Open to a conversation. Kubernetes · GKE · AKS · Istio · Terraform · Go · Python · Helm · Argo Workflows · Argo Rollouts · GitLab CI · Octopus Deploy · GitHub Actions · Prometheus · New Relic · Cloudflare · Workload Identity Federation · MongoDB · PostgreSQL · Redis · Docker · Linux

Location

Gurugram

AI stack

antigravity

claude code

claude

gemini

Work experience

Platform Engineer at dunnhumby, Dec 2024 - Present · 1 yr 10 mos, Gurugram, Haryana, India, Built dhctl, the platform team's Go CLI for cluster connection, PAM requests, cloud resource discovery and per-environment config resolution across Azure and GCP. Replaced the one-off functions every engineer had been writing separately; adopted by all 10 engineers on the team, later moved off shelled-out exec calls onto native Azure and GCP SDKs. Designed the Azure disaster-recovery failover edge for a 5-region multi-tenant platform: flag-driven cutover in code, 5-minute recovery when Cloudflare is unavailable. Ruled out Azure Front Door after TLS-termination mismatch and asymmetric user-defined routes broke standby health probes; shipped on Application Gateway. Contributed to Istio on GKE, including cross-cluster mesh trust for multi-cluster tenants; brought up several of the first cross-cluster connections and wrote the corner-case procedure and docs later tenant migrations followed. Led platform releases across 4 environments — about 20 projects in dev, test and pre-prod, and 80 in production. Contributed to retiring long-lived GCP service-account keys in favour of Workload Identity Federation — AKS pods now enqueue GCP runbook work with short-lived, scope-limited STS credentials instead of a static key in the pod. Co-owned bucket backups for 80+ platform projects, targeting an isolated GCP project and Azure subscription. Fixed defects in the system and in its CI/CD, and replaced per-error ServiceNow tickets with batched alerts.

Senior Engineer at Thoughts2Binary Consulting & Solutions, Aug 2023 - Dec 2024 · 1 yr 5 mos, Gurugram, Haryana, India, Designed secure, scalable Terraform templates for flexible multi-region deployments and serverless architectures. Streamlined deployments 20% faster using AWS CodePipeline for seamless migrations and optimized builds on ECS clusters. Automated MongoDB deployment and autoscaling for unparalleled efficiency and 99% uptime. Implemented robust monitoring and alerting systems for proactive issue identification. Led containerization efforts leveraging ECS and Docker for optimized resource utilization and streamlined application management. Championed knowledge transfer through clear documentation of infrastructure configurations. Optimized costs through efficient capacity planning and strategic deployment of Python automation scripts. Key Achievements: 2x Award Winner (Thoughts2Binary): "Budding Star" (2022) & "Star Performer" (2023) 8x Faster Deployments: Developed a Terraform template for automated environment setup (reduced time from 1 day to 8 minutes). Client Recognition: Praised for both exceptional MongoDB setup and significant cost reduction strategies.

DevOps Engineer at Thoughts2Binary, Apr 2021 - Mar 2023 · 2 yrs, Designed and led the migration off Bitbucket onto AWS CodePipeline and CodeBuild, roughly 100 pipelines end to end: automated test gates, KMS-encrypted artifacts in S3, container images published to ECR. Co-owned a Go resource-tracking system on Lambda: CloudTrail events routed through EventBridge, every resource catalogued with its owner at creation across 5 AWS accounts, so finance could attribute spend per team. Co-owned automating 6 operational runbooks in Python and Bash.

Intern at SSDN Technologies, May 2020 - Aug 2020 · 4 mos, 1. Designed and implemented an auto-scalable architecture for a webapp within a VPC. Used AWS S3 for storage and AWS RDS for database. Created private and public subnets which were associated to Elastic Load Balancer for managing traffic. JumpHost & Web Server on Public Subnet and DB, monitoring Server on Private Subnet. Implemented a provision for health check and monitoring. Implemented a bot control method using AWS WAF for increased security. 2. Setup Continuous Delivery of weather application in docker using Jenkins Declarative pipelines. Pulled the code from GitHub open-source repository, performed compilation, dockerized the app, then pushed the image to registry, and deployed the app

Education

Bachelor of Technology - BTech, Electrical, Electronics and Communications Engineering, Shri Ram Murti Smarak (SRMS) Institutions

XII, Shri Gulab Rai Montessori Sr. Sec School

Public links

github: https://github.com/pratyushsinha05