First seen by Alion on Nov 19, 2026.
You are the engineer our customers work with directly. You take a business problem nobody has written down properly, work out what should actually be built, build it, deploy it into an environment you don't control, and stay with it until it's being used in production every day.
The requirement you're given on day one is usually not the requirement that ships. Working out the difference is the job.
We hire two types of engineers into this role. You need to be strong in one and credible in the other—we don't expect both at depth.
Infrastructure-led. You're strong in cloud, Linux, networking, Kubernetes, and deployment automation. You've debugged a VPC routing problem and a failing Helm release in the same week. You write solid production code and can design a service end-to-end.
Systems-led. You're a strong engineer—you design production systems and reason about data models, concurrency, failure modes, and APIs. You're comfortable in Linux and the cloud and have shipped through Docker and Kubernetes; you just haven't been the person who owns the cluster.
Both need the same thing that isn't negotiable: you can work out what a customer actually needs when nobody has told you.
Key responsibilities:
- Figure out what to build. Work with the customer's technical and business teams to understand the workflow the platform is meant to change. Decide what the deployment actually needs—and say no to the parts that won't move the outcome.
- Design for their reality. Private networks, corporate firewalls, restricted egress, existing databases, enterprise identity, data residency. Design something that works inside those constraints, not around them.
- Build it. Write production code—integrations, adapters, data pipelines, and deployment automation. You're in the codebase, not reviewing someone else's.
- Deploy and operate it. Containers, Kubernetes, CI/CD, networking, observability. Own it through go-live and through the first weeks of real usage, when the interesting failures happen.
- Troubleshoot across every layer. An issue arises as "it's slow. " You find out whether that's the model, the query plan, the network path, or their proxy.
- Scope and estimate. Infrastructure sizing, BOMs, deployment plans, dependencies—and an honest account of what you can't commit to yet.
- Feed the field back into the product. You'll see the same gap at the third customer that you saw at the first. Turning that into a product change, sometimes by building it yourself, is part of the role.
- Lead a small team while staying hands-on. Two to four engineers across concurrent deployments. You stay in the code.
Required qualifications & tech stack:
- 5+ years building and shipping production systems, with real ownership of something customers used
- Strong Linux fundamentals and practical networking: you can reason about DNS, routing, TLS, proxies, and firewalls when something breaks
- Hands-on production experience on AWS or Azure (both is a plus)
- Docker and Kubernetes in production—managed services (EKS / AKS / GKE) count
- Strong in at least one of Python, Go, or Java, and able to work in an unfamiliar one
- Comfortable across REST APIs, relational databases and SQL, CI/CD pipelines, and production debugging
- You can hold a technical conversation with a customer's architect and a difficult one with their security team
- You learn unfamiliar systems quickly, because every deployment contains at least one
Nice-to-haves:
None of these are gates. They're what makes the first few months faster.
DevOps & infrastructure:
- Owning CI/CD end-to-end—GitHub Actions, GitLab CI, Azure DevOps, ArgoCD
- Observability in production: Prometheus, Grafana, OpenTelemetry, ELK; building dashboards and alerts people actually act on
- Air-gapped or restricted-egress deployments—private registries, image mirroring, offline installs
- Cluster-level Kubernetes: ingress, storage classes, autoscaling, resource tuning, upgrades
- Secrets management (Vault, Key Vault, Secrets Manager) and certificate handling
Data and services:
- PostgreSQL, MongoDB, Redis
- Kafka, RabbitMQ, or SQS
- Vector stores and retrieval pipelines
Enterprise and security:
- AD, SAML, OIDC, and SSO integration; RBAC design
- Security reviews, compliance frameworks, VAPT remediation
Other:
- GCP and on-premises deployments
- Having built on LLMs and understanding how model behavior shapes what the product can promise
- Leading small teams without leaving the code
Skills
Docker, Go, Java, Kubernetes, Linux, Python

