https://haribaskar.dev/ AI Engineer building practical LLM systems, and teaching what I learn. en Sat, 10 Oct 2026 00:00:00 GMT https://haribaskar.dev/blog/building-an-agent-from-first-principles https://haribaskar.dev/blog/building-an-agent-from-first-principles Sat, 10 Oct 2026 00:00:00 GMT Build an AI agent from scratch in Python: tools, the observe-decide-act loop, state, planning, error handling, termination and verification. agents llms python https://haribaskar.dev/blog/context-engineering https://haribaskar.dev/blog/context-engineering Fri, 09 Oct 2026 00:00:00 GMT Beyond better prompts: how to design, assemble, optimize and manage the information an LLM needs to solve a task. llms rag agents https://haribaskar.dev/blog/prompt-to-next-token https://haribaskar.dev/blog/prompt-to-next-token Fri, 09 Oct 2026 00:00:00 GMT A deep technical guide to LLM internals, from raw text to token generation, with Python implementations from scratch. llms deep-learning python https://haribaskar.dev/blog/production-ai-caching-architecture https://haribaskar.dev/blog/production-ai-caching-architecture Thu, 08 Oct 2026 00:00:00 GMT Bringing it all together: how to design caching as a first-class part of an AI system, from keys and versions to reliability, cost and rollout. mlops rag agents llms https://haribaskar.dev/blog/ai-cache-observability https://haribaskar.dev/blog/ai-cache-observability Mon, 05 Oct 2026 00:00:00 GMT Measure what the cache actually saves: latency, tokens, calls avoided, cost, ROI, hit quality and freshness, not just the hit rate. mlops evals llms https://haribaskar.dev/blog/ai-cache-invalidation https://haribaskar.dev/blog/ai-cache-invalidation Thu, 01 Oct 2026 00:00:00 GMT Fast is useless if the answer is wrong: TTLs, versioned keys, dependency graphs, events, concurrency and testing for correct AI caches. mlops rag llms https://haribaskar.dev/blog/distributed-ai-caching https://haribaskar.dev/blog/distributed-ai-caching Mon, 28 Sep 2026 00:00:00 GMT Scaling AI caches: Redis Cluster sharding, hot keys, L1/L2 caches, stampedes, invalidation at scale, multi-region, failure handling and cost. mlops llms https://haribaskar.dev/blog/llm-inference-caching https://haribaskar.dev/blog/llm-inference-caching Thu, 24 Sep 2026 00:00:00 GMT KV cache, prefix cache and prompt caching inside LLM inference: how they differ, what they cost in GPU memory, and how to design prompts for reuse. llms mlops deep-learning https://haribaskar.dev/blog/caching-agent-work https://haribaskar.dev/blog/caching-agent-work Mon, 21 Sep 2026 00:00:00 GMT Agent state, plans, tool and API results, workflow steps, checkpoints, idempotent writes and scoped sharing: caching the work an agent does. agents llms mlops https://haribaskar.dev/blog/rag-caching https://haribaskar.dev/blog/rag-caching Thu, 17 Sep 2026 00:00:00 GMT Make retrieval faster, cheaper and smarter: caching queries, embeddings, retrieval, reranking, chunks, context and answers without serving stale data. rag llms mlops https://haribaskar.dev/blog/conversational-ai-caching https://haribaskar.dev/blog/conversational-ai-caching Mon, 14 Sep 2026 00:00:00 GMT Caching for chat: conversation history, sessions, summaries, user memory, context fingerprints and responses, without leaking or going stale. llms agents mlops https://haribaskar.dev/blog/agentic-ai-caching https://haribaskar.dev/blog/agentic-ai-caching Thu, 10 Sep 2026 00:00:00 GMT Caching for agents: tool results, workflow steps, state, checkpoints, MCP tools and resources, idempotency, and keeping it all safe. agents llms mlops https://haribaskar.dev/blog/ai-caching-playbook https://haribaskar.dev/blog/ai-caching-playbook Mon, 07 Sep 2026 00:00:00 GMT KV, prompt, semantic, embedding, retrieval, reranking, tool, API, conversation and agent state caches: what each saves and how to key it safely. llms rag agents mlops https://haribaskar.dev/blog/production-rag https://haribaskar.dev/blog/production-rag Thu, 03 Sep 2026 00:00:00 GMT From demo to dependable: retrieval, reranking, grounding, citations and evaluation for retrieval-augmented generation in production. rag llms mlops