DevOps and AI/ML engineering for teams that run real infrastructure
Blackfir Systems builds delivery pipelines, infrastructure as code, network automation and LLM systems, then hands them over documented and tested so your team can run them without us.
-
Lintyamllint, ansible-lintPassed14 s
-
Render configsdal1-lf-07 to dal1-lf-12 from source of truthPassed9 s
-
Pre-change validationBatfish reachability, 212 flows, 0 regressionsPassed1 m 42 s
-
Canarydal1-lf-07, EVPN routes and MLAG state checkedPassed38 s
-
Deploydal1-lf-08 to dal1-lf-12Running3 of 5
-
Post-change diffBGP and EVPN state, before and afterWaiting
What we build
Four areas of work. Most projects touch two of them, because a model that can't be deployed and a pipeline with nothing worth shipping are the same problem.
Platform and delivery
CI/CD that engineers trust enough to use for every change, and infrastructure defined in code with a reviewable plan before anything is applied.
- Build, test and deploy pipelines
- Terraform and Ansible codebases
- GitOps for Kubernetes workloads
- Runbooks and on-call handover
Network automation
Network changes handled like software changes: a source of truth, generated configs, validation before the push and a state diff after it. Multi-vendor by default.
- Source of truth and config templates
- Pre-change impact checks
- Staged rollouts with automatic rollback
- Migration and cutover tooling
AI and ML systems
LLM applications that answer from your own documents and systems, with evaluation in place so you can tell whether a change made them better or worse.
- Retrieval over internal documentation
- Agents and MCP servers with guardrails
- Self-hosted and local model deployment
- Evaluation sets and regression tests
Security in the pipeline
Controls that run on every commit instead of once a year: policy as code, secrets handling, image and dependency scanning, and Kubernetes hardening.
- Policy checks in CI
- Secrets management rollout
- Cluster and workload hardening
- Firewall policy as code
How an engagement runs
Three shapes of work. Every one ends with something written down that your team keeps.
Assessment
We review the current pipeline, platform or network and deliver written findings ranked by risk and effort, with a plan you can run yourselves or hand back to us.
Build
A defined system delivered in your repositories, through your review process, with tests, documentation and a handover session at the end.
Embedded
Ongoing engineering alongside your team for work that doesn't fit a fixed scope: migrations, on-call improvements, tooling and reviews.
About
Blackfir Systems is an independent engineering consultancy. You work directly with the engineer doing the work.
The practice comes out of more than ten years of network engineering in multi-vendor datacenter and WAN environments, where a bad change takes down something people depend on. That background shapes how we approach automation and AI alike: validate before you push, keep a way back, and write down how it works.
- Network
- Cisco IOS, NX-OS and ACI, Arista, Juniper
- Security
- Palo Alto, Fortinet, F5, Zscaler
- Protocols
- BGP, VXLAN/EVPN, DMVPN, SD-WAN
- Platform
- Terraform, Ansible, Kubernetes, GitOps
- AI and ML
- Python, LLM applications, MCP, local models
Start with an email
Tell us what you're trying to change. You'll get a reply from an engineer within two business days.
info@blackfirsystems.com- What you run today: vendors, cloud, rough size
- What you want to be true in three months
- What has been tried already
- Any deadline or change window driving the timing