https://www.covaratech.com/blog Covaratech is an agentic engineering company. We build, test, deploy and scale AI agents and agentic systems for real business workflows. en Sat, 24 Oct 2026 07:27:10 GMT https://www.covaratech.com/blog/nvidia-minerva https://www.covaratech.com/blog/nvidia-minerva Fri, 23 Oct 2026 00:00:00 GMT Julie A technical look at NVIDIA Minerva, the AIOps platform for AI factory observability: architecture, the knowledge graph, goodput, and how it compares to DCGM. ai-infrastructure observability nvidia sre ai-factory https://www.covaratech.com/blog/cdm-models https://www.covaratech.com/blog/cdm-models Wed, 21 Oct 2026 00:00:00 GMT Julie What are constrained decoding models (CDMs)? How token masking works, the engines compared, real vLLM flags, and why valid output is not correct output. llm-inference structured-output constrained-decoding ai-infrastructure vllm https://www.covaratech.com/blog/deepInfer-inference https://www.covaratech.com/blog/deepInfer-inference Wed, 21 Oct 2026 00:00:00 GMT Julie How DeepSeek DeepInfer works: architecture, parallelism, DSpark, YoCo, EPLB, kernels, benchmarks, and how it compares to vLLM and SGLang. llm-inference deepseek deepinfer ai-infrastructure vllm https://www.covaratech.com/blog/spec-driven-agentic-era https://www.covaratech.com/blog/spec-driven-agentic-era Tue, 13 Oct 2026 00:00:00 GMT Julie A practical and honest look at spec driven development with coding agents: Spec Kit, Kiro, plan mode, when specs pay off, and when they are pure overhead. ai-agents software-engineering spec-driven-development developer-tools ai-coding https://www.covaratech.com/blog/how-to-optimize-kv-cache https://www.covaratech.com/blog/how-to-optimize-kv-cache Mon, 12 Oct 2026 00:00:00 GMT Julie A practical guide to KV cache optimization: memory math, prefix caching, FP8 quantization, CPU and NVMe offloading, and cross instance cache pools with real flags. llm-inference kv-cache vllm sglang ai-infrastructure https://www.covaratech.com/blog/token-economics https://www.covaratech.com/blog/token-economics Mon, 12 Oct 2026 00:00:00 GMT Julie A practical guide to LLM token economics: the cost per token formula, every factor that drives it, and the metrics to track for inference cost visibility. llm-inference finops token-economics ai-infrastructure observability https://www.covaratech.com/blog/zero-trust-mcp-authorization https://www.covaratech.com/blog/zero-trust-mcp-authorization Thu, 01 Oct 2026 00:00:00 GMT Julie What PayPal's zero-trust MCP paper found: per-tool authorization, filtered tool lists, dual-persona auth, and why hiding a tool from an agent is not enough. mcp ai security security ai agents https://www.covaratech.com/blog/agent-orchestration-using-ax https://www.covaratech.com/blog/agent-orchestration-using-ax Mon, 28 Sep 2026 00:00:00 GMT Julie What is Google AX? A deep dive into the open source declarative orchestrator for AI agent workloads, with architecture, primitives, and a full worked example. ai-infrastructure kubernetes ai-agents ax agent-substrate https://www.covaratech.com/blog/plugin4shell-coding-agent-vulnerability https://www.covaratech.com/blog/plugin4shell-coding-agent-vulnerability Tue, 22 Sep 2026 00:00:00 GMT Julie Plugin4Shell breaks SHA pinning in Claude Code, Codex, Copilot, and Gemini CLI. Here's how the exploit works and what we found checking our own stack. ai security devsecops coding agents supply chain security claude code https://www.covaratech.com/blog/what-is-jev https://www.covaratech.com/blog/what-is-jev Mon, 21 Sep 2026 00:00:00 GMT Julie What is TypeSafe Jev? Learn how System One models deliver fast, type safe, calibrated AI decisions and build a daily triage harness in Python. what-is ai-infrastructure automation typesafe jev llm https://www.covaratech.com/blog/what-is-llmd-and-hami https://www.covaratech.com/blog/what-is-llmd-and-hami Thu, 17 Sep 2026 00:00:00 GMT Julie What are llm-d and HAMi? Learn how these Kubernetes projects improve LLM inference, GPU sharing, scheduling, and private AI infrastructure. ai-infrastructure kubernetes llmd hami gpu https://www.covaratech.com/blog/devsecops-for-ai-agents https://www.covaratech.com/blog/devsecops-for-ai-agents Mon, 14 Sep 2026 00:00:00 GMT Julie A practical guide to DevSecOps for AI agents: threat models, Kubernetes isolation, tool authorization, supply chain, runtime security and scaling. devsecops security kubernetes ai infrastructure https://www.covaratech.com/blog/what-is-hugging-face-open-alignment-initiative https://www.covaratech.com/blog/what-is-hugging-face-open-alignment-initiative Sun, 13 Sep 2026 18:00:00 GMT Julie What Hugging Face's Open Alignment Initiative is, why Thomas Wolf is leading it, and how it ties into Anthropic's push for embedded safety evaluators. ai safety alignment open source https://www.covaratech.com/blog/what-is-doclang https://www.covaratech.com/blog/what-is-doclang Sun, 13 Sep 2026 00:00:00 GMT Julie Why PDF, DOCX and Markdown break AI pipelines, and how DocLang, an open document standard, fixes the bottleneck behind hallucinations and bad extraction. document-ai rag https://www.covaratech.com/blog/enterprise-rag-in-production https://www.covaratech.com/blog/enterprise-rag-in-production Sat, 12 Sep 2026 00:00:00 GMT Julie What enterprise RAG actually looks like in production in 2026: hybrid retrieval, permission enforcement, multi-agent architecture, and the honest failure modes. rag enterprise ai multi-agent systems ai infrastructure https://www.covaratech.com/blog/how-to-build-good-mcp-tools https://www.covaratech.com/blog/how-to-build-good-mcp-tools Sat, 12 Sep 2026 00:00:00 GMT Julie Lessons from running 50+ MCP servers in production: schema design, authentication, transport security, error handling, SSRF, SQL safety and testing. mcp ai tooling ai infrastructure security https://www.covaratech.com/blog/what-is-agent-memory https://www.covaratech.com/blog/what-is-agent-memory Fri, 11 Sep 2026 00:00:00 GMT Julie How AI agent memory works in production: the types of memory, the architecture for retaining and retrieving context, and how to build one that scales. agents-memory https://www.covaratech.com/blog/success-without-correctness https://www.covaratech.com/blog/success-without-correctness Mon, 07 Sep 2026 00:00:00 GMT Julie Three silent failures in a Docling and Tesseract OCR ingestion pipeline, and why 'it didn't throw' is not the same as 'it worked'. document processing ocr rag ai infrastructure https://www.covaratech.com/blog/open-knowledge-format https://www.covaratech.com/blog/open-knowledge-format Sat, 05 Sep 2026 00:00:00 GMT Julie Open Knowledge Format v0.2 is Google's attempt to make organizational knowledge portable, machine-readable, and trustworthy for AI agents. ai agents open knowledge format okf agentic ai enterprise ai ai infrastructure knowledge management https://www.covaratech.com/blog/six-billion-requests-later https://www.covaratech.com/blog/six-billion-requests-later Thu, 20 Aug 2026 00:00:00 GMT Julie A full year of unsampled production LLM traffic, 6.12 billion requests across 9,174 models, and what it changes about caching, routing and capacity. llm inference