/tags/long/ Recent content in Long on Condor Writing Hugo en Tue, 07 Apr 2026 00:00:00 +0000 /posts/llmdev2026/ Tue, 07 Apr 2026 00:00:00 +0000 /posts/llmdev2026/ <div class="ox-hugo-toc toc"> <div class="heading">Table of Contents</div> <ul> <li><a href="/posts/llmdev2026/#introduction">Introduction</a></li> <li><a href="/posts/llmdev2026/#microservices">Microservices</a></li> <li><a href="/posts/llmdev2026/#resources-and-constraints">Resources and Constraints</a> <ul> <li><a href="/posts/llmdev2026/#latency-hierarchy">Latency Hierarchy</a></li> <li><a href="/posts/llmdev2026/#device">Device</a></li> <li><a href="/posts/llmdev2026/#network">Network</a></li> <li><a href="/posts/llmdev2026/#data-and-storage">Data &amp; Storage</a></li> </ul> </li> <li><a href="/posts/llmdev2026/#testing">Testing</a> <ul> <li><a href="/posts/llmdev2026/#testing-layers">Testing Layers</a></li> <li><a href="/posts/llmdev2026/#iteration-speed-is-development-output">Iteration Speed is Development Output</a></li> <li><a href="/posts/llmdev2026/#effective-docker-design">Effective Docker Design</a></li> </ul> </li> <li><a href="/posts/llmdev2026/#involving-llms">Involving LLMs</a> <ul> <li><a href="/posts/llmdev2026/#working-with-models-via-ooda-loops">Working with Models via OODA Loops</a></li> <li><a href="/posts/llmdev2026/#model-selection">Model Selection</a></li> <li><a href="/posts/llmdev2026/#claude">Claude</a></li> <li><a href="/posts/llmdev2026/#gpt">GPT</a></li> <li><a href="/posts/llmdev2026/#open-models">Open Models</a></li> <li><a href="/posts/llmdev2026/#model-collaboration">Model Collaboration</a></li> <li><a href="/posts/llmdev2026/#effective-collaboration-tools">Effective Collaboration Tools</a></li> </ul> </li> <li><a href="/posts/llmdev2026/#the-human-in-the-machine">The Human in the Machine</a></li> <li><a href="/posts/llmdev2026/#wrap-up">Wrap Up</a></li> </ul> </div> <!--endtoc--> <p>This is a talk I gave at a company on-site turned into an article. Expect it to be lengthy and less suitable for a non-technical audience after the introduction. The article is intended for new developer onboarding, detailing a range of infrastructure considerations, and a guide to working with LLMs for enterprise development.</p> /posts/gtc2026/ Thu, 19 Mar 2026 00:00:00 +0000 /posts/gtc2026/ <div class="ox-hugo-toc toc"> <div class="heading">Table of Contents</div> <ul> <li><a href="/posts/gtc2026/#gtc-highlights">GTC Highlights</a></li> <li><a href="/posts/gtc2026/#talks">Talks</a> <ul> <li><a href="/posts/gtc2026/#scaling-out-and-across-networking-innovations-for-giga-scale-ai-systems-s81561">Scaling Out and Across: Networking Innovations for Giga-Scale AI Systems [S81561]</a></li> <li><a href="/posts/gtc2026/#build-a-high-performance-research-cluster-s81731">Build a High-Performance Research Cluster [S81731]</a></li> <li><a href="/posts/gtc2026/#inside-nvidia-dgx-ai-factory-accelerating-networking-for-ai-across-cloud-core-and-edge-s81856">Inside NVIDIA DGX AI Factory: Accelerating Networking for AI Across Cloud, Core, and Edge [S81856]</a></li> <li><a href="/posts/gtc2026/#achieve-truly-serverless-gpus-with-libfuse-criu-and-cuda-checkpoint-s81424">Achieve Truly Serverless GPUs With libfuse, CRIU, and CUDA-Checkpoint [S81424]</a></li> <li><a href="/posts/gtc2026/#accelerate-cloud-platforms-for-the-next-era-of-ai-s81788">Accelerate Cloud Platforms for the Next Era of AI [S81788]</a></li> <li><a href="/posts/gtc2026/#how-we-scaled-kimi-k2-dot-5-s81695">How We Scaled Kimi K2.5 [S81695]</a></li> <li><a href="/posts/gtc2026/#vllm-in-2026-architectural-challenges-and-performance-optimizations-s82059">vLLM in 2026: Architectural Challenges and Performance Optimizations [S82059]</a></li> <li><a href="/posts/gtc2026/#general-networking">General Networking</a></li> <li><a href="/posts/gtc2026/#building-measuring-and-using-ai-scientists-s81694">Building, Measuring, and Using AI Scientists [S81694]</a></li> <li><a href="/posts/gtc2026/#an-ai-driven-autonomous-lab-of-the-future-for-chemistry-and-materials-science-s81790">An AI-Driven Autonomous Lab of the Future for Chemistry and Materials Science [S81790]</a></li> <li><a href="/posts/gtc2026/#the-state-of-open-source-ai-s81791">The State of Open Source AI [S81791]</a></li> <li><a href="/posts/gtc2026/#a-new-paradigm-verifiable-ai-s81489">A New Paradigm: Verifiable AI [S81489]</a></li> <li><a href="/posts/gtc2026/#optimize-kv-caches-for-llm-inference-dynamo-kvbm-flexkv-lmcache-s82033">Optimize KV Caches for LLM Inference: Dynamo KVBM, FlexKV, LMCache [S82033]</a></li> <li><a href="/posts/gtc2026/#mlops-202-from-models-to-production-ai-systems-at-scale-s81662">MLOps 202: From Models to Production AI Systems at Scale [S81662]</a></li> <li><a href="/posts/gtc2026/#science-and-engineering-with-ai-physics-and-kit-cae-s81781">Science and Engineering With AI Physics and Kit-CAE [S81781]</a></li> </ul> </li> <li><a href="/posts/gtc2026/#conclusion">Conclusion</a></li> </ul> </div> <!--endtoc--> <h2 id="gtc-highlights">GTC Highlights</h2> <p>GTC 2026 is in the books. I did my best to make the rounds. Below is a lengthy list of talks and conversations I had on a wide range of relevant subject material. I figured I&rsquo;d give a quick summary for those less interested in the details.</p>