https://ashimsharma10.github.io/blog Software engineer sharing projects, notes, and guides on ML infrastructure. en-us sharmaashim00@gmail.com (Ashim Sharma) sharmaashim00@gmail.com (Ashim Sharma) Sun, 02 Aug 2026 00:00:00 GMT https://ashimsharma10.github.io/blog/ai-engineers-swiss-knife https://ashimsharma10.github.io/blog/ai-engineers-swiss-knife Forty production questions on agentic systems, answered plainly: evals, guardrails, quality measurement, observability, security, and token economics. Sun, 02 Aug 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) llmagentsevalsguardrailsguide https://ashimsharma10.github.io/blog/architecture-of-memory-kv-cache https://ashimsharma10.github.io/blog/architecture-of-memory-kv-cache Modern LLM inference is memory-bound, and the KV cache is the reason. This report walks through why the cache exists, how large it gets, the attention changes that shrink it (GQA, DeepSeek MLA), how engines like vLLM and SGLang manage it, how quantization and eviction compress it, and how disaggregated serving moves it between machines. Sat, 05 Sep 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) kv-cachellminferenceattentiongpu https://ashimsharma10.github.io/blog/evaluation-engineering https://ashimsharma10.github.io/blog/evaluation-engineering How to tell whether an LLM system works, and how to keep it working: test cases, graders, judges, groundedness, CI gates, live signals, traces, latency, schemas, injection defense, and PII masking. Thu, 13 Aug 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) evalsobservabilityguardrailsllmagents https://ashimsharma10.github.io/blog/from-prompting-to-loops https://ashimsharma10.github.io/blog/from-prompting-to-loops Why AI work moved from single prompts to loops, the main types of loops, and examples of goal-driven loops in practice. Mon, 20 Jul 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) agentsllmloop-engineeringai https://ashimsharma10.github.io/blog/graph-engineering-for-agentic-ai https://ashimsharma10.github.io/blog/graph-engineering-for-agentic-ai What graph engineering is, how it differs from loops and workflow engines, when to build one, a refund approval worked through end to end, and the failure modes to watch for. Tue, 28 Jul 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) agentsaillmgraph-engineeringmulti-agent https://ashimsharma10.github.io/blog/harness-context-loop-engineering https://ashimsharma10.github.io/blog/harness-context-loop-engineering A practical field guide to the systems that wrap an LLM and turn it into a reliable agent — with patterns, worked examples, real deployments, and where the field is heading (mid-2026). Sat, 13 Jun 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) agentsllmharness-engineeringloop-engineeringcontext-engineeringai https://ashimsharma10.github.io/blog/mixture-of-experts https://ashimsharma10.github.io/blog/mixture-of-experts Sixteen questions about sparse models, each one the thing you would ask after hearing the last answer: what the router really is, why experts do not learn topics, why sparsity stops paying once you batch, and what it costs to serve one. Mon, 17 Aug 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) moellmarchitectureinferencetraining https://ashimsharma10.github.io/blog/rlvr-and-the-experience-era https://ashimsharma10.github.io/blog/rlvr-and-the-experience-era Post-training has moved from imitating human labels to learning from verifiable outcomes. This write-up covers the RLVR objective, GRPO and the choice of KL penalty, advantage collapse and the methods that recover the lost gradient, reward hacking and noisy verifiers, outcome versus process reward models, the production data flywheel that turns failures into weight updates, long chain-of-thought and latent reasoning, and the OpenRLHF stack that runs it all. Sun, 06 Sep 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) rlvrreinforcement-learninggrpollmcontinual-learningreasoning https://ashimsharma10.github.io/blog/self-improving-agents https://ashimsharma10.github.io/blog/self-improving-agents An agent that gets better on its own has to change something about itself, and something has to check the change. This post walks the six things it can change, the checker that decides whether any of it worked, the ceiling it runs into, the ways it breaks, and the loop teams actually run in production. Thu, 13 Aug 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) agentsaillmself-improving-agentscontinual-learningloop-engineering https://ashimsharma10.github.io/blog/vllm-how-a-token-gets-served https://ashimsharma10.github.io/blog/vllm-how-a-token-gets-served What happens inside a serving engine: why the GPU fills up with something other than the model, what paging the cache buys you, why reading a prompt and writing an answer fight each other, and which knobs settle the fight. Thu, 13 Aug 2026 00:00:00 GMT sharmaashim00@gmail.com (Ashim Sharma) vllmllminferencegpukv-cache