Sheet 00/Cover/Samskrutha Mruthyunjaya Swamy
I'm Samskrutha, a senior DevOps and platform engineer in Berlin. I design, build and run the cloud infrastructure that software lives on: Kubernetes, Terraform, delivery pipelines, and the observability to see all of it working.
Twelve years in software, the last seven on the platform side, across automotive, finance, digital health and an AI startup. Below is what I can do, how I work, and where I've done it.
Fig. 1 The life of a change. Click a station to see what I can do there.
I can set up repositories so the right things happen the moment code lands: branch protection, code owners, pre-commit hooks, secret scanning, automated dependency updates, and review flows a team will actually keep.
GitHub · GitLab · Bitbucket · Dependabot · pre-commit
I can build pipelines that produce the same artefact every time: slim container images with sensible caching, vulnerability scanning inside the pipeline, versioned Helm charts and Terraform modules, and tests that gate the merge instead of decorating it.
Docker · GitHub Actions · GitLab CI · Jenkins · Helm · Black Duck
I can design deploys that are boring on purpose: GitOps with Argo CD or Flux, selective deploys that only touch what changed, releases that can't trample each other, rolling and blue/green rollouts, and a rollback that is one command.
Argo CD · Flux · Terraform · Helm · Kubernetes · ECS
I can run production on Kubernetes or ECS with autoscaling, load balancing and fault tolerance, databases that are backed up and provably restorable, networking that is written down, and zero-trust access for the people who need to get in.
Kubernetes (CKA) · EKS · GKE · ECS · RDS / Postgres · Cloudflare Access · VPN
I can stand up observability from nothing: metrics, logs and traces with consistent tagging, dashboards that tell the truth, alerts that only fire when someone should move, on-call rotations and an incident process, and the cost controls that keep all of it affordable.
Datadog · Prometheus · Grafana · Loki · Elastic · New Relic
I can turn what went wrong into something durable: blameless postmortems, runbooks people can follow at three in the morning, RFCs that set direction, walkthroughs that teach the whole team, and handovers that don't need me in the room.
Postmortems · Runbooks · RFCs · Knowledge-sharing · Onboarding
Sheet 01/Capabilities
A few ways I can help your team. Every one of these is something I've done in production, not something I've read about. Open any card for the specifics.
Networking, compute, databases, IAM, secrets, tagging and cost attribution, all as code from day one. I write reusable Terraform modules so the second environment is a pull request, not a project.
Done it: greenfield AWS platforms for automotive projects, a migration to a sovereign Google Cloud for a digital-health company, and bringing a hand-built startup cloud fully under Terraform.
EKS, GKE or self-managed. Helm charts, GitOps with Argo CD or Flux, autoscaling, high availability, upgrades without downtime, and the guardrails (resource limits, network policies, admission rules) that keep a shared cluster civil.
Done it: fault-tolerant clusters for high-traffic applications, GKE provisioned from Terraform with Argo CD wired in, and the ELK and Prometheus stacks running inside the clusters they watch.
GitHub Actions, GitLab CI or Jenkins. Selective deploys that only ship what changed, serialised releases, security scanning in the pipeline, environment promotion, and rollbacks that take one command.
Done it: designed delivery pipelines from scratch for teams of five to fifty, including the deploy mechanism a startup still runs on today.
Datadog, Prometheus and Grafana, Elastic or Loki. Consistent service tagging, dashboards that are actually correct, alerts tied to who should move, SLOs, and the written guide and walkthroughs so every team can debug its own services.
Done it: built a startup's observability platform from nothing, ran the trainings, and kept the bill in check without losing visibility.
On-call design, incident command, clear communication while things are on fire, and the postmortem, runbook and alert fixes afterwards so the same thing is caught earlier next time. I'm comfortable owning the database and infrastructure side of an outage.
Done it: the standing incident commander for platform and database outages at a startup, including restoring production databases from backup during a live incident.
Owning controls, collecting evidence before the auditor asks, working through cloud security findings one by one, and the access side: SSO, zero-trust access to internal systems, key rotation, WAF, logging retention.
Done it: owned the infrastructure controls and evidence for a startup's first SOC 2 and ISO 27001 certifications, and wrote the incident-management process.
Tagging standards that make spend attributable per service and team, right-sizing, log and metric hygiene, regular cost reviews across cloud and SaaS vendors, and negotiating the contracts myself when it helps.
Done it: made per-service spend visible for the first time at a startup and cut the observability bill substantially while keeping every signal that mattered.
Reliable pipelines, data contracts with versioning and compatibility rules, freshness and integrity checks that run before anything is published, and a catalogue so consumers can find what they need. Python and TypeScript when there's code to write, SQL always.
Done it: co-leading the move from datasets to governed data products on a startup's data engineering team.
I've led a DevOps team of five, mentored engineers, onboarded new hires from day one and handed over whole areas cleanly. A Certified Scrum Product Owner, so I'm comfortable with roadmaps, stakeholders and saying no with a reason.
Done it: led the DevOps team for automotive projects in Wolfsburg, and later onboarded and handed over to my successor at a startup.
Sheet 02/How I work
Things I've come to believe after enough deploys, enough pages, and enough mornings after. Written down so I'm held to them.
Writing the proposal is half the job. The other half is the module nobody thanks you for and the boring tag on every service. Direction without plumbing is a slide deck.
You can't cut what you can't see, and you shouldn't cut what you don't trust. Get the dashboards right first. Then the bill comes down on its own, and nothing goes dark on the way.
Every monitor should answer two questions: who moves, and what do they do first? If neither has an answer, it's a graph, not an alert. Graphs are fine. Just don't page people with them.
The first time, you're allowed to improvise. The second time, someone should be reading your notes instead of calling you. If they still have to call you, the notes weren't finished.
A platform that only I understand is a liability with a dashboard. The real finish line is the walkthrough, the written guide, and the day the next person owns it without needing me in the room.
A lot of this job is being reachable when it matters, answering fast, and never making anyone feel small for asking. If people are afraid to ping the platform team, the platform team hears about problems last.
Sheet 03/Experience
Three countries on three continents, and a slow drift from writing the application to owning the ground it runs on. Click a stop on the map, or use the arrows.
2008 – 2012 · Mysore, India
Bachelor of Engineering in Computer Science from Visvesvaraya Technological University. Four years of algorithms, operating systems and networks, and the first hint that I liked the systems underneath more than the screens on top.
2012 – 2014 · Kalamazoo, Michigan
Master of Science in Computer Science at Western Michigan University. Distributed systems, databases, a lot of snow, and the habit of reading the whole stack before trusting any part of it.
2014 – 2019 · Dallas, Texas
Senior software engineer with HCL, building Java and React applications for clients in finance and travel. Five years of shipping to real deadlines, tuning the SQL, and writing the tests first. It is where I noticed I cared most about how software got to production, and started volunteering for that part.
2019 – 2024 · Wolfsburg, Germany
Lead DevOps engineer at Hexad, on Volkswagen Group projects. This is where I became a platform person: fault-tolerant AWS platforms on Kubernetes, everything in Terraform, the Jenkins pipelines with GitOps and security scanning built in, and the Elastic logging stack and dashboards the teams relied on. Real traffic, real on-call, and a team to lead.
2024 – now · Berlin, Germany
First a senior DevOps engineer at Perfood, migrating a digital-health platform to a sovereign Google Cloud with Terraform, GKE, Argo CD and a Prometheus and Grafana stack, where compliance isn't optional. Since 2025, platform, infrastructure and observability engineer at Bluefish AI, a Berlin startup, serving every engineering team, and lately a data engineer there too.
Sheet 04/Bill of materials
Honest labels: daily means I live in it, often means I reach for it most months, when needed means I've shipped with it and can again. Tools change, and I'll happily learn yours.
| Layer | Materials |
|---|---|
| Cloud | AWSGoogle CloudAzure |
| Orchestration | KubernetesECSHelmArgo CDFluxDocker |
| Infrastructure as code | TerraformCloudFormationAnsible |
| Delivery | GitHub ActionsGitLab CIJenkinsGitOps |
| Observability | DatadogPrometheusGrafanaLokiElasticNew Relic |
| Data | Postgres / RDSSQLStep FunctionsData contractsRedshift |
| Security & compliance | SOC 2ISO 27001IAM / SSOWAFZero-trust access |
| Languages | PythonShellTypeScriptJava |
Sheet 05/Contact
I'm in Berlin and I like hard infrastructure problems. If you need someone who can build the platform and then run it, write me a note. I read every one.
Off the clock I'm with my wife and our two-year-old, or listening to a tech podcast while building something that doesn't need to exist yet.