Pratilipi
Website:
pratilipi.com
Job details:
About the rolePratilipi is building a generational company in storytelling, and infrastructure is core to that bet. We're looking for an Infrastructure Engineer — a hands-on role embedded in a high-growth product team using AWS, CI/CD, our self-managed data layer, and the operational substrate for the AI workloads now landing in production. You'll help us self-host our own models and take the infrastructure multi-region as we expand globally.
Expect a lot of breadth, with depth in a few things that matter. You won't be a specialist hiding behind a narrow remit — you'll move across AWS, CI/CD, databases, CDN, and AI infra, and go deep where it counts.
What you'll do- Own our AWS infrastructure through Terraform — ECS on EC2, VPCs, IAM, autoscaling, rolling deploys. Clickops-free console as the target.
- Build Jenkins pipelines that are fast, safe, and self-serve. Be in the deploy paths, not just the platform.
- Drive cost optimisation.
- Own media delivery at scale — Cloudflare CDN, image/video/audio transcoding, format/device delivery — tuned for latency, cache hit ratio, and egress cost.
- Operate the self-managed data layer with service teams — RDS MySQL, MongoDB, Valkey, plus managed MSK and ScyllaDB Cloud. Lead migrations end-to-end (e.g. the ongoing Redis → Valkey).
- Stand up the operational substrate for AI in production — GPU capacity, inference gateways, key rotation, rate limits, cost-per-request visibility. Build observability for LLM agents: traces, token/cost accounting, eval hooks, alerts on silent regressions.
- Partner with the AI/ML team on self-hosting models — capacity planning, vLLM / TGI serving, canary rollouts, and cost/latency trade-offs vs managed providers.
- Own reliability across services and infra — SLOs, alerting, incident response, blameless postmortems. Move teams from firefighting to proactive reliability.
- Once you have the context, add the guardrails — IaC checks, deploy gates, paved paths — that make the right thing the easy thing and quietly remove whole classes of human error.
What we're looking for
- 4–6 years in DevOps, SRE, or infrastructure — production systems at meaningful scale, with ownership beyond tickets.
- A problem solver with high agency. You reason from first principles, don't wait to be told, and dig in rather than deflect — whether it's a developer stuck on Terraform or an ML engineer asking for GPUs.
- Strong AWS hands-on — ECS on EC2, VPCs, IAM, multi-AZ design — and proficiency with Terraform (you write modules others reuse).
- Jenkins in production plus a real sense of developer experience in CI/CD — you've looked at deploy-time metrics and changed them.
- Python, Ansible, and shell — you automate work rather than repeat it.
- Operated databases in production — at least some of RDS MySQL, MongoDB, Redis/Valkey. Done migrations, failovers, and perf tuning, not just provisioning. Working familiarity with Kafka and Cassandra/ScyllaDB.
- Strong grasp of Linux internals and networking fundamentals (TCP/IP, DNS, TLS, load balancing) and common failure modes.
- Some hands-on exposure to LLM-based systems — inference endpoints, agent tracing, token/cost, or evals in CI. Not a researcher; you should reason clearly about latency, cost, and failure modes of LLM workloads.
- Comfortable with cloud security fundamentals: IAM least-privilege, secrets management, network segmentation. Prior exposure to ISO 27001 / DPDP-style controls is a plus.
Tech stack
AWS · ECS (on EC2) · Terraform · Jenkins · Ansible · Python · Cloudflare · Prometheus · Grafana · InfluxDB · RDS MySQL · MongoDB · Valkey · MSK · ScyllaDB Cloud · growing: GPU inference · vLLM / TGI · OpenTelemetry GenAI · Langfuse-style tracing · multi-region AWS
Security & Data Handling
All employees are expected to handle sensitive data responsibly in compliance with the DPDP Act, ISO-27001:2022, and Pratilipi's internal security policies — ensuring data privacy, confidentiality, and NDA obligations at all times.
Click on Apply to know more.
