Tag
9 essays filed under LLM Engineering.
How to secure agent orchestration without trusting proxies: attested runtimes, per-agent sandboxes, deterministic state, wipe protocols, and end-to-end networki
Read essayKV caches aren’t harmless memory. Use provenance-aware eviction and metering to prevent corrupted reasoning, leakage, and runaway spend—with concrete designs.
Read essayAuto-mode LLMs shift risk to teams. Build human-in-the-loop contracts—UI gates, runtime checks, provenance, and metering—to keep control and costs predictable.
Read essayStop shipping agents on JS/SaaS glue. Compile agents to native, typed runtimes to cut risk, cost, and gatekeeping while improving auditability and control.
Read essayA practical metering model for agents: bill by work units, not tokens. Prevent leaks, rein in costs, and keep provenance using an enforceable boundary.
Read essayTreat repos, APIs, and CI as agent interfaces. Concrete patterns to cut hallucinations, debug AI-generated code, and reduce AI coding costs.
Read essayMake the terminal the control plane for agent orchestration: an auditable, cost-effective, developer-centric way to run enterprise agent workflows at scale.
Read essayRising SaaS AI costs? An engineering-first playbook to shift workloads to a staged, hybrid self-host strategy without disrupting the business.
Read essayHow to build agent memory you can trust: cryptographic provenance, scoped capability tokens, and SLOs for enterprise-ready, auditable recall.
Read essay