SRE - Platform Engineer

📋 Description

• • Own the beating heart of Aqemia’s drug-discovery engine. You will architect, build, and harden the cloud-native platform that lets 65 chemists, physicists, and ML researchers push 10⁵ molecules per day through quantum-inspired simulations and generative-AI pipelines.
• • Guarantee 99.9 % uptime for mission-critical services that predict binding affinity, enumerate novel scaffolds, and orchestrate terabyte-scale molecular dynamics—because every minute of downtime delays a potential therapy for cancer, inflammation, or CNS disorders.
• • Codify everything. Write Terraform/Terragrunt modules that spin up multi-AZ Kubernetes clusters on AWS, wire IAM least-privilege policies, and provision GPU node groups that scale from 0 to 500 A100s in under five minutes—then open-source the modules internally so every squad can reproduce your stack with one CLI command.
• • Design bullet-proof CI/CD. Extend GitHub Actions workflows with matrix builds that test 40 micro-services in parallel, integrate ArgoCD for GitOps promotion across dev, staging, and production, and cut deployment time from 25 min to <5 min while adding canary and blue-green strategies that protect live experiments.
• • Make scientists’ lives friction-free. Build self-service portals where chemists click once to launch containerized Jupyter notebooks pre-loaded with 2 TB of proprietary assay data, or spin up transient Ray clusters for hyper-parameter sweeps—no YAML wrestling required.
• • Define, measure, and evangelize SLOs. Instrument FastAPI services with OpenTelemetry, pipe traces to Grafana Tempo, and create SLIs that track “molecules evaluated per hour” and “model convergence time.” Run weekly error-budget reviews with ML teams to balance velocity and reliability.
• • Automate incident response. Create PagerDuty playbooks, Terraform-managed runbooks, and Slack-driven ChatOps commands that roll back bad releases, restart failed Airflow DAGs, or drain misbehaving GPU nodes—cutting MTTR from 40 min to <10 min.
• • Optimize for cost and carbon. Implement Karpenter-based auto-scaling policies that save 35 % on EC2 spend, schedule non-urgent batch jobs on Spot GPUs, and publish monthly dashboards that show CO₂ saved per molecule discovered.
• • Secure the entire supply chain. Integrate Trivy scanning in CI, sign container images with Cosign, enforce OPA policies in Kubernetes admission controllers, and run chaos-engineering drills that prove secrets stay sealed even when etcd misbehaves.
• • Influence the technical roadmap. Sit at the table with the CTO, Head of ML, and Head of Chemistry to translate “we need to screen 1 M molecules in 24 h” into concrete infra epics, capacity plans, and quarterly OKRs.
• • Coach and uplift peers. Run internal workshops on Terraform testing, mentor junior engineers through pair-programming, and write decision records (ADRs) that turn tribal knowledge into living documentation.
• • Celebrate real-world impact. When your platform ships a feature that shaves two weeks off lead-optimization timelines, join the all-hands to hear chemists explain how those two weeks may translate into patients getting life-saving drugs sooner.