Soufiane Sejjari — Writing https://soufianesejjari.me/en/blog Notes on LLM agent memory, RAG, and building AI products. en Modeling People as First-Class Entities in Agent Memory https://soufianesejjari.me/en/blog/person-entity-modeling-memory https://soufianesejjari.me/en/blog/person-entity-modeling-memory Flat memory can't answer relational questions like 'how has my relationship with X been trending?'. The entity-layer design behind iRemember: alias resolution across languages, evolving profiles, sentiment timelines, and privacy by construction. Sat, 12 Sep 2026 00:00:00 GMT LLM Memory Knowledge Graphs Compressing Conversation History Without Losing What Matters https://soufianesejjari.me/en/blog/memory-compression-strategies https://soufianesejjari.me/en/blog/memory-compression-strategies Naive summarization destroys exactly the details users ask about later. The compression strategy behind iRemember: episode-vs-fact split, progressive oldest-first gating, temporal anchor preservation, and a probe-question evaluation method. Sat, 05 Sep 2026 00:00:00 GMT LLM Memory RAG Inside a LangGraph Memory Pipeline: From Message to Memory https://soufianesejjari.me/en/blog/langgraph-memory-pipeline https://soufianesejjari.me/en/blog/langgraph-memory-pipeline A teardown of the extraction pipeline behind iRemember: why LangGraph, how a chat message becomes structured memory candidates, deduplication, async processing, and the schema decisions that make retrieval work. Sat, 29 Aug 2026 00:00:00 GMT LLM Memory LangGraph Architecture LLM Agent Memory: The Complete Engineering Guide https://soufianesejjari.me/en/blog/llm-agent-memory-guide https://soufianesejjari.me/en/blog/llm-agent-memory-guide Everything required to build production memory for LLM agents: the four memory operations, hybrid retrieval scoring, write control, history compression, person-entity modeling, and how 2026 benchmarks actually compare. Sat, 22 Aug 2026 00:00:00 GMT LLM Memory RAG Agents Memory Is the Missing Layer of LLM Applications https://soufianesejjari.me/en/blog/memory-is-the-missing-layer https://soufianesejjari.me/en/blog/memory-is-the-missing-layer Context windows keep growing, but they still forget. Why persistent, structured memory — not longer contexts — is what makes LLM assistants actually useful. Sun, 12 Jul 2026 00:00:00 GMT LLM Memory RAG What Building a Multitenant ERP Taught Me About SaaS Architecture https://soufianesejjari.me/en/blog/multitenant-lessons https://soufianesejjari.me/en/blog/multitenant-lessons Tenant isolation sounds like a database decision. It's actually an application-wide discipline. Lessons from shipping a multitenant ERP. Sun, 03 May 2026 00:00:00 GMT Architecture SaaS Spring Boot Notes on NLP for Moroccan Darija https://soufianesejjari.me/en/blog/darija-nlp-notes https://soufianesejjari.me/en/blog/darija-nlp-notes Darija breaks every assumption standard NLP pipelines make: no standardized orthography, code-switching mid-sentence, and almost no labeled data. Field notes from building sentiment and transcription tools. Wed, 18 Feb 2026 00:00:00 GMT NLP Darija Research