Analyses et guides
Articles
Analyses approfondies sur l'infrastructure IA, le MLOps, les clusters GPU et le développement avec des agents.
Building AI Agents with NebulaDB: A Technical Guide from the Labs
AlloyDB AI ideas rebuilt for an open-source database: filtered semantic SQL, hybrid search, step-by-step RAG, access control in the query, determinis…
21 min readLire la suite
Sizing and Securing Offline-First AI Workstations for Law Firms: GPU Memory Maths, Isolation Options and a Staging Checklist
VRAM and KV-cache arithmetic per tier, a review of a real first quote, TPM-bound encryption with unattended restart, offline patching and model updat…
13 min readLire la suite
Inside a 32-GPU HGX B300 AI Factory: Compal SGX30-2 Nodes, a Rail-Optimised 800G Fabric and the Workflows It Runs
Bill of materials, node architecture, four networks, the software stack, model-sizing maths and the acceptance tests we run before handover
14 min readLire la suite
Technote: OpsAPI Single Sign-On for Java, PHP, Python, Rust and Go Apps, with Zero-Trust Admin Access
The request path, the sign-in contract with real Java, PHP, Python, Rust and Go code, configuration, zero-trust admin access, testing and lessons lea…
23 min readLire la suite
An IP Allowlist Isn't a Login: VPN Edges, DNS Failover and OpsAPI SSO for Lower Environments
AllowedIPs and hub NAT for every failover edge, the OpsAPI hosted sign-in contract, silent SSO, and a test plan
15 min readLire la suite
An IP Allowlist Is Not a VPN Check: Zero Trust Access Profiles with wslvpn and WSLProxy
Zero Trust access profiles with wslvpn and WSLProxy: CIDR, identity, fail-closed behaviour and debug headers
15 min readLire la suite
Gemini 4 Argon in Practice: API Changes, Costs and Ten Long-Horizon Workflows
Workstation engineering guide: 1M unified window, long decode, thinking levels, Argon + Flash orchestration, cost model, safety, and ten worked workf…
16 min readLire la suite
