Our on-premises text seat had run at its chat template's highest reasoning-effort setting since the day it was deployed, because nothing in the stack ever set the value. Measured across 94 benchmark items per setting, that default scored 84.6 on our flat composite against 93.4 at the middle setting, while spending 81 percent more completion tokens; disabling reasoning entirely was worse than both, so the reasoning was doing work and its length was the defect. The dimensions measuring raw reasoning were already at ceiling and could not improve. Everything that moved measures whether the model commits: tool selection, abstention, multi-step execution, grounding. Abstention rose from 66.7 to 100.0, meaning maximum effort had reasoned the model into answering questions it should have refused. Because the settings are inserted sentences rather than compute budgets, we then reproduced the top setting as a hand-written expert-persona prompt and watched tool selection fall to the same 62.5 it scored at maximum effort. We give the method, the non-termination failure at the top of the dial, and an explicit account of what these numbers will not carry.
ReadA production inference engine reported the same aggregate throughput at one concurrent request as at sixty-four. It was not batching at all: every request past the first was queueing, and the single benchmark point we had been quoting could not see the difference. Sweeping the concurrency ladder corrected a stored throughput figure by 3.2× and moved a model comparison from a reported ~7× to 17.0×, or 33.9× per GPU, since NVIDIA's Nemotron 3.5 Lightning sustained 5,613 tokens per second across 64 concurrent requests on a single Blackwell card, at sub-three-second p95, and had not saturated when our ladder ran out. We give the diagnostic signature (flat aggregate, linearly climbing p95), the four-configuration sweep that exposed it, a quality result that survived repetition and one that did not, and why a recorded performance number belongs to a configuration rather than to a model.
ReadFor five months a nightly backup reported Complete over a zero-byte archive. An alert rule had been structurally incapable of firing for its entire life. A dashboard rendered missing telemetry as health, and 37 of 49 deployment scripts could silently ship stale monitoring state. We name the failure class, the false green, map its mechanisms in four families from a 24-day production audit, show that verification tooling itself reproduces the class it hunts, and describe the falsification harness now grading all 279 of our alerting controls continuously: registry, structural liveness, mutation-proven firing tests, and end-to-end canaries. Monitoring correctness is a reachability property, not a syntax property.
ReadAn agentic coding CLI ballooned to 178.5 GiB of resident memory during a routine fleet upgrade and froze a production workstation. The kernel killed fourteen innocent processes before touching the culprit. A forensic walk through the OOM killer's victim ordering, why swapless machines turn leaks into freezes, the four-layer guard we deployed the same afternoon, and the upstream report we filed instead of posting screenshots.
ReadOur GPU compute tier is powered off most of the time by design. That makes any persistent state, or any hard scheduling pin, that lands on it a silent failure waiting to happen. Here is how we audited every hard dependency, relocated all state onto the storage cluster, and then proved resilience the only way that counts: by cutting the power.
ReadWe set out to design a self-healing v2 and found the opposite of a greenfield: four generations of self-healing stacked in sediment, maximum architectural sophistication, near-zero closed-loop coverage. The fix was not a fifth system: it was naming the missing shape, and then discovering the real blocker was signal quality, not architecture.
ReadAn agent's memory is its most dangerous component: poison it quietly and every downstream decision inherits the lie. We built a memory substrate on one governing principle (Markdown is the source of truth, Postgres is a rebuildable cache) and then learned the hard way that our retrieval eval had been measuring the wrong thing all along.
ReadGoogle shipped its first open-weights text-diffusion language model: a different generation paradigm that paints a whole canvas of tokens in parallel instead of writing left to right. We benchmarked it within days of release, and a 3.8B-active model beat a 128B production model on hard mathematics.
ReadCan a 550-billion-parameter open model be served on Blackwell B200 and stand up to a frontier closed model on real agentic tool use? The serving path is a gauntlet of era-specific gotchas, and the most important result in the whole campaign is the one we published, retracted, and corrected.
ReadA benign-looking storage warning led to an uncomfortable discovery: every production database on the cluster had its data on an erasure-coded pool, paying a partial-stripe read-modify-write penalty on every single fsync. The disks were fine. The pool was wrong, by about 17×.
ReadA routine fabric test found one inter-node link running at 29% of capacity, not broken, just quietly starved of PCIe lanes. Chasing that thread all the way down took us from a hidden slot-bifurcation trap to NCCL collective ceilings to a per-lane optical asymmetry that error-correction was silently hiding.
ReadA sovereign GPU fleet emits a dozen overlapping signals (GPU box health, Kubernetes workloads, distributed storage, metrics, out-of-band power) and operators were jumping between tools to read them. The engineering story here is what it takes to turn an internal cockpit into something safe to expose: config-driven topology, read-only-by-default gates around every dangerous action, and secret-scanning in CI.
ReadFive frontier open-weight mixture-of-experts models (MiniMax-M2.7, GLM-5.1, DeepSeek-V3.2, Qwen3-235B and Qwen3.5-397B) served on the same three-Blackwell cluster under llama.cpp TCP RPC and graded on one mechanical 30-prompt suite at matched budget. A hybrid Mamba2+transformer leads at 96.7%, and default thinking modes turn out to be a methodology trap.
ReadCan a single-operator workstation cluster with no NVLink and no InfiniBand serve a 228-billion-parameter mixture-of-experts reasoning model at conversational speed? Part 3 distributes MiniMax-M2.7 at Q8_0, 226 GiB across three Blackwell GPUs, under llama.cpp over plain TCP RPC, and measures the no-optimisation floor.
ReadDoes a workstation-class Blackwell cluster extend beyond language-model inference? Part 2 retargets the same two-node, three-GPU cluster to a full GPU-accelerated click-through-rate pipeline (Parquet ingest, high-cardinality shuffle, target encoding, and multi-GPU XGBoost) on roughly 750 GB of amalgamated system RAM over RAPIDS and Dask-CUDA.
ReadCan frontier open-weight language models be served on workstation-class NVIDIA Blackwell GPUs interconnected by commodity RDMA Ethernet, without NVLink or InfiniBand? Part 1 of an ongoing PureTensor research programme brings up the cluster, characterises the fabric, verifies GPUDirect RDMA, and demonstrates first cross-node inference of Llama 3.1 405B.
ReadThe third paper in the CMAS research series, and the artefact the methodology was built to produce. Cosmology is the conduit; the question is whether structured adversarial interaction between frontier AI models can produce research-grade technical output on an open problem where no ground truth exists. The paper below is what came out.
ReadCan a vague metaphysical idea, lacking any mathematical grounding, be refined into a falsifiable research programme through structured adversarial interaction between frontier AI models? A case study in speculative cosmology.
ReadCan structured adversarial interaction between independently trained LLMs produce insights that exceed any single model's capability? We present a theoretical framework and initial case study in black hole information theory.
ReadAn analysis of how autonomous code-generation agents are compressing multi-month development cycles into days, and what the structural implications are for the software engineering profession.
ReadOver 80% of enterprise AI initiatives fail before reaching production. The bottleneck is not model capability but context architecture. A technical analysis of the emerging discipline of context engineering.
ReadAdversaries are stockpiling encrypted data today, betting that quantum computers will break current encryption within a decade. What this means for organizations running sensitive AI workloads.
ReadThe global chip supply chain depends on a handful of irreplaceable facilities, materials, and monopolies. A technical assessment of concentration risk and its implications for AI compute sovereignty.
ReadAI progress is increasingly constrained not by algorithms but by the physical infrastructure required to run them. An analysis of power, cooling, chip lifespan, and supply chain fragility.
ReadAI agents are not improving enterprise SaaS. They are replacing it. An analysis of the structural threat to per-seat licensing models and the emerging 'service-as-software' paradigm.
ReadAI automation increases corporate efficiency but simultaneously reduces consumer purchasing power. An examination of the macroeconomic feedback loop that could constrain AI adoption.
ReadGoogle's Willow processor demonstrated a 13,000x speedup over classical computation. What this actually means for AI infrastructure, and what it does not mean for encryption.
Read