Platform / DevOps Engineer, Client Embedded Pod (Kubernetes / Observability / Data Infrastructure)

Experience: 5+ years, Mid-Senior Level 

 About

Langoor, A born in digital era digital marketing agency is challenging/changing the very way marketers make sense of the digital disruption and navigate their way to Impact Business x Marketing x Brand Performance.

Going beyond the elementary impact of digital in terms of search, social and storytelling, marketers need to look at their customer journeys in the digital domains and focus their action based on insights of digital behaviour and contextise their marketing mix strategies and action.

Driving Indian Brands into the Digital First Marketing Era with our expertise in Data x Technology x Strategic Thinking x Creative Edge delivering agile Marketing Transformations & Brand Performance. 

About the Role

You will join an embedded pod building the platform foundations that product and engineering teams run on, directly inside a client environment on live production systems. This is not a maintenance role. In a single quarter you could be hardening Kubernetes cluster lifecycle and capacity, tuning ClickHouse for high-cardinality observability, and standing up or scaling Milvus for a retrieval or agent workload. The pod moves fast and the client expects production-grade infrastructure, not one-off scripts.

What we expect from you

•    You write clean, production-ready infrastructure and platform code, and you ship it fast. You do not need heavy process to move.

•    You pick up unfamiliar stacks and tools quickly and without much hand holding – Kubernetes operators, GitOps, telemetry pipelines, columnar and vector databases.

•    You already use modern AI tools to design, debug, and operate systems, and can show what that has done to your output.

•    You are comfortable with ambiguity and switch context across infrastructure domains – cluster work one week, data-store tuning the next – without dropping reliability.

•    You can hold a technical conversation directly with a client stakeholder, not just with your own team.

•    You consistently turn platform work into measurable outcomes: developer velocity, cost control, and uptime.

Must-have skills

•    Kubernetes in production: cluster lifecycle management, Helm charts and operators, capacity planning, and multi-tenant platform design

•    GitOps workflows for infrastructure and application delivery (ArgoCD, Flux, or similar)

•    ClickHouse or a comparable columnar store: schema design, ingestion pipelines, and query performance tuning for high-cardinality observability data (logs, traces, metrics, wide events)

•    Vector databases in production – Milvus or similar (Pinecone, Weaviate, pgvector) – operating data planes for retrieval and agent workloads

•    Telemetry pipeline design spanning logs, traces, and metrics

•    Demonstrated use of AI tools in your engineering workflow, and awareness of how to apply AI capabilities to platform operations

•    Git and version control workflows in a team setting (branching, PRs, code review)

Good-to-have skills

•    Writing or extending Kubernetes operators and custom controllers

•    Experience with complementary observability tooling (Prometheus, Grafana, OpenTelemetry)

•    Cloud platform experience – AWS, Azure, or GCP

•    CI/CD pipeline ownership, not just exposure

•    Cost optimization or capacity forecasting for infrastructure spend

•    Exposure to building AI agents or working with LLM APIs

•    Comfortable writing infrastructure tests and validation as part of your normal workflow

•    Experience working in fast, iterative cycles (agile or otherwise) with shifting priorities

•    Prior experience working directly embedded with a client rather than purely internal teams

What this role is not

•    This is not a role for someone who prefers a fixed scope and stable domain. The problem domain changes – cluster hardening one week, ClickHouse tuning the next, Milvus scaling after that.

•    This is not an entry-level supervised role. You will be expected to operate independently within a few weeks of starting.