Builds
Systems we built, and still run.
These are written by the engineers who built them. They include the parts that broke, because a case study without a failure section is a brochure.
GeekyAnts internal platform
Standing up on-prem LLM inference next to the private cloud
Serving open-weight models on owned GPUs with vLLM — sizing against KV cache, multi-tenant scheduling, and an OpenAI-compatible gateway in front.
- vLLM
- Continuous batching and prefix caching
- No egress
- Prompts and documents stay in-network
- OpenAI API
- Drop-in gateway for existing code
- Quota
- Per-team token accounting
bolt.sh platform
Building our own anycast CDN on a /24 we own
We acquired IPv4 space, got an ASN, announced the same prefix from POPs on AWS and Vultr with BIRD, and ran the edge ourselves. Here is what it took and what broke.
- /24
- IPv4 block we own and announce
- 2 clouds
- AWS and Vultr in one anycast fabric
- RPKI
- Signed ROAs on every announcement
- BIRD 2
- eBGP sessions we operate ourselves
GeekyAnts internal platform
Running OpenStack as an internal private cloud
Self-service compute for engineering teams on hardware we operate — Nova, Ceph, Terraform tenancy and the operational lessons that only show up after month three.
- Self-serve
- Teams provision via API and quota
- Ceph
- Replicated block and S3-compatible object
- Terraform
- Same workflow as public cloud
- Hybrid
- Routed to public cloud regions
GeekyAnts IT
Zero-touch device management across a distributed workforce
Mac, Windows and Android fleets on Kandji, Intune and SureMDM — one compliance definition, three platforms, device posture gating production access.
- 3 platforms
- macOS, Windows and Android under one policy
- Zero-touch
- Devices ship direct to the employee
- Posture
- Compliance state gates production access
- Same day
- Offboarding revokes and wipes
Client engagements are covered by confidentiality agreements. We can walk through anonymised architectures and introduce you to references on a call — ask when you speak to us.
Next step
Tell us what breaks at 3am.
A 30-minute call with the engineers who would do the work — not a sales desk. We will tell you whether this is a bolt.sh problem or something you can fix in-house.