Software Engineer - Infrastructure

📋 Description

• • Own the beating heart of Coram AI’s hybrid infrastructure. You’ll architect, build, and harden the systems that keep hundreds of thousands of edge devices and cloud services in perfect sync, ensuring every camera, sensor, and AI agent delivers real-time intelligence without missing a frame.
• • Design and maintain a cloud-native backbone on AWS (EKS, RDS, S3, MSK, IAM) while simultaneously crafting the custom edge stack that runs on our Yocto-based IoT devices. This is not a “configure once” role—you’ll iterate daily, shipping IaC (Pulumi/Terraform) that provisions clusters, networks, and security policies in minutes, not weeks.
• • Build bullet-proof CI/CD and automation pipelines that push code from laptop to production in under 15 minutes, covering container builds, Helm charts, OTA updates for edge devices, and canary deploys across multiple regions. Your pipelines will be the gold standard that the rest of the engineering org relies on.
• • Instrument end-to-end observability: deploy Prometheus, Grafana, Loki, and custom exporters so that every metric, trace, and log from cloud APIs to ARM boxes in a parking garage is queryable in a single pane of glass. When a device drops offline at 2 a.m., you’ll know why before the customer does.
• • Create zero-touch device lifecycle tooling—from factory flashing and secure boot to remote wipe and certificate rotation—so our field fleet can scale to millions without human intervention. You’ll also harden the supply chain (SBOMs, CVE scanning, signed artifacts) to meet SOC 2 and HIPAA requirements.
• • Optimize for developer velocity: build self-service environments, ephemeral namespaces, and golden-path templates that let ML engineers spin up GPU clusters or firmware engineers test OTA deltas without opening tickets. Your work will cut “time to first commit” for new hires to under an hour.
• • Collaborate tightly with ML, firmware, and product teams to co-design features like on-device model hot-swapping, bandwidth-aware clip upload, and edge-side event buffering. You’ll translate product requirements into resilient architectures and realistic SLAs (99.9 % uptime, <100 ms P99 inference latency).
• • Champion security-first thinking: enforce IAM least privilege, network micro-segmentation, secrets management (KMS, Vault), and continuous compliance scanning. When auditors arrive, your runbooks and evidence packages will make the audit feel like a formality.
• • Participate in an on-call rotation that is humane, blameless, and constantly improving. Every post-mortem you run becomes a blueprint for eliminating an entire class of failure, not just a single incident.
• • Shape the technical culture of a fast-growing startup. Whether it’s debating the merits of eBPF vs sidecar proxies or presenting a lunch-and-learn on reproducible builds, your voice will influence how Coram AI scales for the next decade.