Lead Cloud Architect

Role: Lead Cloud Architect
Experience: 8+ years (3+ leading cloud/platform architecture)
Location: Delhi / Kolkata
Employment Type: Full-Time

About the Role

Aria builds enterprise agentic-AI products, and this role owns the cloud architecture they run on. We are looking for an
architect — not a generalist cloud engineer — who has designed the foundations under multi-agent and GenAI platforms
and knows Kubernetes and AWS deeply (strong GCP or Azure backgrounds are equally welcome; the architectural
thinking matters more than the badge).
This is a hands-on leadership role: you set the architecture, standards and roadmap, and you build alongside the team.
You will work daily with agent engineering, data science and product, and your decisions shape how the platform scales,
how it is secured and observed, and how it eventually ships into customer environments.

Key Responsibilities

What you will architect

• The Kubernetes platform under a multi-service agentic system — cluster design, networking and ingress, autoscaling, workload isolation, and safe sandboxing for code that AI agents generate and execute.
• Agent-aware infrastructure — service-to-service and agent-to-agent communication patterns: identity and per- service credentials, secrets management, least-privilege access, and discovery.
• LLM serving and gateway infrastructure — managed model services (e.g. AWS Bedrock), model routing/tiering, and token-level usage and cost metering across tenants and features.
• GitOps-driven delivery — everything as code: Terraform/IaC, declarative CD (ArgoCD or similar), ephemeral per-PR environments, repeatable environment lifecycle from dev to production.
• Data infrastructure for AI workloads — object storage, queues and eventing, and Postgres-family stores serving RAG and analytical workloads at the billions-of-rows scale.
• Observability and reliability — metrics, logs and traces with alerting and SLOs; DR, backup, and incident management as first-class design concerns.
• Security and cost as architecture inputs — IAM design, network segmentation, encryption; FinOps guardrails built into the platform rather than bolted on.

Where we are heading — what you will drive

Today our platform runs as a managed service on AWS. The roadmap this role leads:

1. Platform-as-a-Service — evolve the platform so it can be deployed into a customer’s own cloud account (AWS first,
   then Azure/GCP) as a repeatable, versioned, one-command installation with clean upgrade paths.
2. Cloud-agnostic core — abstractions over storage, model access and infrastructure primitives so the same platform
   runs on any major cloud (and eventually on-prem) without redesign.
3. Fleet observability — a centralized monitoring, alerting and telemetry setup giving proactive visibility across every
   environment we or our customers run, including usage and cost telemetry per deployment.
4. Productized release engineering — versioned releases, automated environment provisioning and teardown, and
   the operational tooling that lets a small team run many environments safely.
   The architect we hire will own the technical shape of all four.

Required Skills & Qualifications

• Education: Bachelor’s or Master’s in Computer Science, Engineering or a related field — or equivalent demonstrated experience.
• 8+ years in cloud/platform engineering, with 3+ years owning architecture decisions in production.
• Deep AWS: EKS, IAM, VPC and networking, load balancing, secrets management. (Comparable depth on GKE/AKS with working AWS knowledge also qualifies.)
• Kubernetes beyond deployment: ingress and service networking, autoscaling, Helm, security contexts and workload isolation, cluster upgrades and multi-environment operation.
• Infrastructure as code (Terraform or equivalent) and GitOps-based delivery (ArgoCD, Flux or similar) as the default way of working — not an aspiration.
• Security architecture: IAM and least-privilege design, secrets lifecycle, network segmentation, encryption in transit and at rest.
• Production observability: Prometheus/Grafana, ELK/OpenSearch or equivalents — designing the setup, not just consuming dashboards.
• Strong Linux, networking and DNS fundamentals; solid Python or Bash.
• A track record of running production: HA and DR design, incident leadership, root-cause discipline.

Must-have: Client Exposure and Leadership

• Client exposure: Client visits; architecture and security reviews.
• Leadership & management: Manages Cloud/DevOps; resource management and admin.

Preferred Skills

• Infrastructure for GenAI/agentic workloads: model gateways, inference scaling, vector stores and RAG pipelines, GPU scheduling, token/cost controls.
• Working familiarity with agent interoperability protocols (MCP, A2A) and what they demand from infrastructure — identity, discovery, long-running sessions.
• Sandboxing untrusted or AI-generated code on Kubernetes (user namespaces, seccomp, gVisor or similar isolation primitives).
• Experience packaging software for customer-operated environments — multi-tenant SaaS evolving to single- tenant/PaaS, installers, upgrade management.
• Distributed data systems in production: Postgres/Aurora at scale, Kafka/SQS-class messaging, Redis.
• FinOps practice; relevant certifications (AWS Solutions Architect Professional or equivalent) are a plus, not a requirement.

What We Look For

• An architect who stays hands-on: comfortable defending a design in review and then implementing it.
• Judgment over dogma — picks boring technology where it wins, new technology where it earns its place.
• Automation-first instincts: eliminates repetitive operational work rather than staffing it.
• Fluent working alongside AI-assisted engineering — uses the tools, understands what they produce, and holds the quality bar on what ships.
• Clear communication with engineers, leadership and customers alike.

Interview Process

Technical screen → architecture deep-dive (you walk us through a platform you designed) → hands-on design session →
leadership conversation.