Lead DevOps & Platform Engineer

Lead DevOps & Platform Engineer
Bangalore, India · Full-time · 10+ years

About NuPlay AI

NuPlay AI (formerly Nurix AI) builds enterprise AI agents for voice, chat and workflows, plus the platform that runs them and keeps improving them. More than 150 enterprises run NuPlay agents in production, including Aditya Birla Capital, Myntra, Groww, Tata Digital and super.money. Our voice agents handle more than 5 million call minutes a month.

NuPlay was founded by Mukesh Bansal (Myntra, Cult.fit) and has raised over $42M from Accel, General Catalyst and Prosus. In July 2026 we acquired Verloop.io to extend our chat and voice reach. We hold ISO 27001, ISO 42001 and SOC 2 Type 2 certifications.

The role

You will own the platform every NuPlay product runs on: cloud, Kubernetes, GPU inference, observability, incident response and access control. First, you will lead our move from AWS to Azure without dropping a call. Alongside that, you will build the observability, incident management and access controls that enterprise customers expect.

This is a hands-on lead role. You write the Terraform, the tooling and the runbooks, review service designs with our engineers, and set the standard the team follows.

What you’ll own

Our platform runs voice agents (cascade and speech-to-speech), chat agents, deep RAG and enterprise search, user and agent memory, a self-hosted multi-tenant auth service, agent-building frameworks, post-call analytics and our own fine-tuned LLMs.

• Cloud architecture and the Azure migration:

  • Design landing zones, multi-region networking (hub-and-spoke, Private Link, NAT, layered trust zones) and HA/DR against defined RPO and RTO targets.

  • Lead the AWS → Azure migration and run both clouds safely in between.

  • Keep everything as code in Terraform.

• Kubernetes and the AI runtime:

  • Run EKS and AKS clusters for microservices, real-time voice pipelines and GPU serving of our fine-tuned LLMs.

  • Autoscale for call peaks, hold the voice latency budget, and track GPU cost per call minute.

• Observability and incident management:

  • Build observability across the application, data and model layers: metrics, logs, traces, and SLOs per product and per tenant.

  • Set up on-call, severity levels, incident command, customer status updates and blameless RCAs.

  • Feed alerts into our agents’ auto-heal loops.

• Tenant isolation, security and access:

  • Ship each sub-product as shared multi-tenant SaaS, as a dedicated tenant stack, or into a customer’s own cloud, all from one codebase and one pipeline.

  • Run RBAC and just-in-time access for internal staff across clusters, databases, vector stores, logs and call recordings, with full audit trails.

  • Own the infrastructure controls and evidence for SOC 2 and ISO audits.

  • Keep the auth service highly available.

• Delivery and developer platform:

  • Own CI/CD and GitOps across all sub-products, with safe rollouts and fast rollbacks.

  • Build internal tooling in code.

  • Review service designs so they scale and fail well.

• Team leadership:

  • Lead and grow the DevOps team: hire, mentor, and set the standards for infrastructure and on-call.

  • Act as the bridge between product engineering and infrastructure.

What we’re looking for

• 8–12 years in total. You started as a backend engineer shipping production services in Python, Go, Java or Node.js, then led an infrastructure, platform or SRE charter for at least 3 years.

• You have designed, not just operated, production cloud on AWS, all in Terraform or an equivalent.

• Hands-on Azure experience (AKS, VNets, Entra ID, Private Link), or you have led a cloud-to-cloud migration.

• Deep production Kubernetes: multi-cluster operations, networking and ingress, service mesh, stateful workloads, GitOps (Argo CD or Flux), and autoscaling with HPA, KEDA or Karpenter.

• You have built observability and incident processes others relied on: OpenTelemetry, Prometheus and Grafana or Datadog, SLOs and error budgets, on-call rotations and RCAs.

• You have run multi-tenant SaaS for enterprise customers: IAM, RBAC, secrets management, network isolation and audit logging.

• At least one year running AI products in production.

Nice to have

• Real-time voice infrastructure: SIP, WebRTC, telephony carriers and per-hop latency budgets.

• GPU inference for LLMs: vLLM, TensorRT-LLM or Triton, GPU scheduling and cost control.

• Search, vector and streaming systems in production: OpenSearch, Weaviate or pgvector, Kafka.

• Identity and compliance: Keycloak, OIDC and SAML; SOC 2, ISO 27001, ISO 42001, HIPAA, GDPR or India’s DPDP Act audits.

• CKA, CKS or Azure Solutions Architect Expert certification.

What success looks like

• First 90 days: On-call, severity levels and the incident process are live. SLO dashboards cover voice latency and availability for every product. The Azure landing zone and migration plan are signed off.

• By 6 months: The first products serve production traffic on Azure. Internal access runs through RBAC and just-in-time grants, with audit logs. A dedicated-tenant deployment ships from the standard pipeline.

• By 12 months: The migration is complete with no customer-visible downtime. DR has been tested against its RPO and RTO. Infrastructure evidence for SOC 2 and ISO audits is collected automatically.

Our interview process

1. Recruiter screening (20 min): your career path from backend engineering to platform leadership.

2. Architecture round (90 min): design NuPlay’s voice platform on Azure at scale.

3. Incident deep dive (60 min): walk through a live incident scenario and one you have led.

4. Hands-on round (90 min): review a Terraform module and a service PR, then extend a small tool.

5. Leadership conversation (45 min): with our CTO or a founder.