Engineering notes — Muhammad Imaz Anwar

Engineering notes — Muhammad Imaz Anwar https://imazanwar.com/blog/ Source-backed technical writing on system design, production engineering, algorithms, AI systems, and research. en Database isolation anomalies, MVCC, and PostgreSQL SSI https://imazanwar.com/blog/database-isolation-anomalies/ https://imazanwar.com/blog/database-isolation-anomalies/ Tue, 28 Jul 2026 00:00:00 GMT Databases A schedule-level guide to snapshot isolation, write skew, serialization graphs, and how PostgreSQL SSI prevents non-serializable commits. HNSW vector search from graph intuition to production tuning https://imazanwar.com/blog/hnsw-vector-search-production-tuning/ https://imazanwar.com/blog/hnsw-vector-search-production-tuning/ Tue, 28 Jul 2026 00:00:00 GMT Search systems A technical guide to HNSW construction, search, M, efConstruction, efSearch, recall-latency-memory trade-offs, filtered search, and reproducible benchmarking. Idempotency and retries in distributed APIs https://imazanwar.com/blog/idempotency-retries-distributed-apis/ https://imazanwar.com/blog/idempotency-retries-distributed-apis/ Tue, 28 Jul 2026 00:00:00 GMT Distributed systems Design retry-safe APIs with idempotency keys, deduplication records, transactional outboxes, bounded exponential backoff, and jitter. Production RAG beyond the demo https://imazanwar.com/blog/production-rag-beyond-the-demo/ https://imazanwar.com/blog/production-rag-beyond-the-demo/ Tue, 28 Jul 2026 00:00:00 GMT RAG systems A practical checklist for turning a promising retrieval-augmented generation prototype into a system you can observe, evaluate, and improve. Designing a distributed rate limiter that fails predictably https://imazanwar.com/blog/designing-distributed-rate-limiter/ https://imazanwar.com/blog/designing-distributed-rate-limiter/ Mon, 27 Jul 2026 00:00:00 GMT System design A system design deep dive into rate-limiting algorithms, Redis data models, atomicity, hot keys, multi-region trade-offs, and failure behavior. The monotonic stack pattern beyond memorized LeetCode solutions https://imazanwar.com/blog/monotonic-stack-pattern/ https://imazanwar.com/blog/monotonic-stack-pattern/ Sat, 25 Jul 2026 00:00:00 GMT Algorithms Derive the monotonic stack from first principles, trace next-greater-element invariants, analyze complexity, and recognize when the pattern does not fit. Debugging a production latency spike without guessing https://imazanwar.com/blog/debugging-production-latency-spike/ https://imazanwar.com/blog/debugging-production-latency-spike/ Thu, 23 Jul 2026 00:00:00 GMT Production engineering Diagnose production latency across queues, application code, databases, caches, and dependencies using percentiles, traces, and saturation signals. AI agents need boring engineering https://imazanwar.com/blog/agents-need-boring-engineering/ https://imazanwar.com/blog/agents-need-boring-engineering/ Tue, 21 Jul 2026 00:00:00 GMT Agent workflows Reliable agent workflows depend less on autonomy theatre and more on explicit tools, state, validation, permissions, and recovery paths. Building an LLM evaluation system that survives production https://imazanwar.com/blog/llm-evaluation-production/ https://imazanwar.com/blog/llm-evaluation-production/ Sun, 19 Jul 2026 00:00:00 GMT AI & LLM systems Design an LLM evaluation stack with task-specific datasets, deterministic checks, model grading, human review, online signals, and regression gates. Attention Is All You Need — engineering notes from the Transformer paper https://imazanwar.com/blog/transformer-paper-engineering-notes/ https://imazanwar.com/blog/transformer-paper-engineering-notes/ Fri, 17 Jul 2026 00:00:00 GMT Research notes Transformer engineering from attention equations and tensor shapes to online softmax, FlashAttention, KV caches, and production kernel dispatch.