Blog
Technical articles on agentic AI, generative AI, scalable infrastructure, and modern web development, written from real-world engineering experience.
Evaluating agents without self-deception: harnesses, contamination, and cost per success
How to evaluate long-running agents through verifiable state, procedural checks, repeated trials, comparable infrastructure, and costs that include failures.
agentic-ai · agent-evaluation · evals · agent-harness · benchmark-contamination · reliability · cost-per-successAgents that write to production: authorization, idempotency, and reconciliation
An execution design for tools with real effects: identity outside the prompt, immutable actions, concurrency control, and recovery when a timeout leaves the result uncertain.
agentic-ai · tool-execution · mcp · authorization · idempotency · distributed-systems · reconciliationRecursive agents: external context, budgets, and evidence that survives delegation
From Recursive Language Models to Recursive Agent Harnesses: designing delegation over large corpora with evidence contracts, global limits, and an evaluation that justifies the complexity.
agentic-ai · recursive-agents · context-engineering · rlm · agent-harness · provenance · inference-budgetsRAG Pipeline Architecture: Chunking Strategies, Hybrid Search, Reranking, and Evaluation Frameworks
A production-focused guide to building Retrieval-Augmented Generation pipelines, from document ingestion and chunking strategies to hybrid search, cross-encoder reranking, and systematic evaluation with RAGAS and custom metrics.
rag · retrieval-augmented-generation · chunking · reranking · hybrid-search · embeddings · llm · evaluation · ragas · ai-agentsAgentic AI Architecture Patterns: Building Autonomous Systems That Scale
A deep technical breakdown of the architecture patterns behind production-grade agentic AI systems, from single-agent loops to multi-agent orchestration, tool calling, and memory management.
agentic-ai · architecture · llm · vercel · ai-sdkTLA+ for Formal Verification: Catching Distributed Systems Bugs Before Writing a Single Line of Code
Modeling distributed systems with temporal logic, running the TLC model checker, and a real-world example of how TLA+ found a bug in a Kafka consumer group rebalancing protocol that testing would never have caught.
tla-plus · formal-verification · distributed-systems · model-checking · temporal-logic · kafka · consensusAgentic AI Architecture Patterns: Tool-Use Orchestration and Multi-Agent Coordination
An analysis of the architecture patterns behind agentic AI systems, from tool-use orchestration to multi-agent coordination, with production-tested approaches and tradeoffs.
agentic-ai · architecture · llm · ai-agents · tool-use · multi-agentWebCodecs API: Hardware-Accelerated Video Encoding and Decoding in the Browser
Low-level video encoding and decoding without [FFmpeg.wasm](https://ffmpegwasm.netlify.app/), VideoEncoder, VideoDecoder, AV1 encoding, and building a real-time video effects pipeline entirely in the browser.
webcodecs · video-processing · av1 · browser-api · video-encoder · hardware-acceleration · webworkersService Mesh Evolution: Istio Ambient Mesh, Linkerd, and Sidecar-Less Architectures
Evaluating sidecar-less data planes: ztunnel, the L4/L7 processing split, mTLS without sidecars, and latency overhead measurements comparing Istio ambient vs sidecar mode vs Linkerd.
service-mesh · istio · linkerd · ambient-mesh · ztunnel · kubernetes · mtls · sidecarDrizzle ORM: Type-Safe SQL for PostgreSQL, Migrations, Prepared Statements, and Prisma Comparison
SQL-like syntax that's actually type-safe, zero runtime overhead, edge runtime support, and an honest comparison with Prisma and Kysely.
drizzle · orm · postgresql · type-safe-sql · prisma · kysely · edge-runtime · migrationsTLS 1.3 + Post-Quantum: ML-KEM (Kyber) Hybrid Key Exchange in the Real World
Inside the X25519Kyber768 hybrid handshake, how Chrome and Firefox implement post-quantum TLS, certificate transparency implications, and what you need to change today.
tls · post-quantum · kyber · ml-kem · cryptography · x25519 · certificate-transparencyModel Context Protocol (MCP): The Missing Standardization Layer for AI-Tool Integration
An examination of MCP's architecture, transport layers, resource/tool/prompt primitives, and how to build custom MCP servers that function reliably in production.
mcp · ai-integration · llm-tooling · anthropic · protocol-design · json-rpcGPU Programming: CUDA vs ROCm vs Vulkan Compute, Warp Scheduling, Shared Memory, and Matrix Multiplication
Warp divergence, shared memory bank conflicts, occupancy tuning, and an optimized matrix multiplication kernel with benchmarks across NVIDIA, AMD, and Vulkan Compute.
cuda · rocm · vulkan-compute · gpu-programming · matrix-multiplication · warp-scheduling · shared-memorySpeculation Rules API: Prerendering Pages Before Navigation and Measured Core Web Vitals Impact
A technical examination of the Speculation Rules API for prerender and prefetch, confidence-based speculation strategies, Chrome's implementation details, and measurable Core Web Vitals improvements including INP optimization.
speculation-rules · prerender · prefetch · core-web-vitals · inp · web-performance · chrome · navigationRISC-V + Rust Firmware: Building an IoT Sensor Node with Embassy-rs on the ESP32-C6
A practical guide to writing no_std Rust firmware for RISC-V targets using embassy-rs async runtime, HAL abstractions, and a real IoT project with the ESP32-C6 running WiFi 6 and BLE 5.
risc-v · rust · embedded · firmware · embassy-rs · esp32-c6 · iot · no-std · hal · async-embeddedClickHouse: Columnar OLAP for Real-Time Analytics on 10 Billion Rows
An architecture guide to ClickHouse for real-time analytics: MergeTree engine internals, materialized views for pre-aggregation, query optimization on 10B+ row tables, and comparative analysis with DuckDB and Druid.
clickhouse · olap · analytics · columnar-database · mergetree · materialized-views · real-time-analytics · duckdb · druidTemporal.io: Replacing Message Queues with Durable Execution for Complex Orchestration
An analysis of Temporal's durable execution model, workflow determinism constraints, activity retries, saga patterns, and why durable execution supersedes orchestration built on top of message queues.
temporal · durable-execution · workflow-engine · saga-pattern · orchestration · distributed-systems · message-queues · backend-architectureSPDK + NVMe: Building a User-Space Storage Engine That Hits 10M IOPS
A technical walkthrough of SPDK's user-space NVMe drivers, polled I/O, zero-copy DMA, and building a storage engine that sustains 10 million IOPS on commodity hardware.
spdk · nvme · storage · performance · zero-copy · dma · io-performance · systems-programming · linux-kernelHomomorphic Encryption in Practice: TFHE, Concrete-ML, and ML Inference on Encrypted Data
A practical guide to TFHE, Concrete-ML by Zama, computing on encrypted data, real ML inference on ciphertext, and the honest performance reality of homomorphic encryption.
homomorphic-encryption · tfhe · concrete-ml · zama · privacy · encrypted-computationEvent Sourcing + CQRS: Event Store Design, Projections, Snapshots, and When NOT to Use It
Production lessons from building event-sourced systems, event store design, projection rebuilds, snapshot strategies, saga orchestration, and an honest assessment of when event sourcing makes things worse.
event-sourcing · cqrs · event-store · projections · sagas · architecture · dddContent Layer API, Server Islands, and View Transitions in Astro 5
An analysis of Astro 5's content layer API, server islands architecture, and view transitions, with production benchmarks comparing performance against Next.js for content-heavy sites.
astro · view-transitions · content-collections · server-islands · hybrid-rendering · nextjs · frontend-architectureSupabase Realtime: Building Multiplayer Features Without Running a Single WebSocket Server
Presence tracking, broadcast channels, Postgres CDC, and scaling WebSocket connections, a practical guide to building multiplayer features with Supabase Realtime.
supabase · realtime · websockets · postgres · presence · multiplayereBPF and XDP: High-Performance Packet Processing and DDoS Mitigation at Line Rate
High-performance packet processing with eBPF and XDP, writing XDP programs in C, AF_XDP sockets, DDoS mitigation at line rate, and peeking inside Cilium's dataplane.
ebpf · xdp · networking · ddos · cilium · linux-kernel · packet-processingCloudflare Durable Objects: Building a Stateful Game Server at the Edge
Stateful edge computing with Cloudflare Durable Objects, WebSocket coordination, the actor model at the edge, and a multiplayer game server architecture running in 300+ locations.
cloudflare · durable-objects · edge-computing · websockets · actor-model · workersDifferential Privacy and Federated Learning: HIPAA-Compliant Healthcare ML Pipeline Architecture
Epsilon budgets, noise mechanisms, secure aggregation, and PySyft in practice, a federated learning pipeline architecture for medical imaging capable of passing HIPAA audits.
differential-privacy · federated-learning · pysyft · healthcare-ml · privacy · secure-aggregationStructured Concurrency in Java 21: Virtual Threads and Server Architecture
An examination of Project Loom's structured concurrency, virtual threads, scoped values, and StructuredTaskScope, and how they fundamentally alter JVM server architecture.
java · virtual-threads · project-loom · structured-concurrency · jvm · scoped-valuesMojo as a Python Superset: Performance on Production ML Workloads
Benchmarking Mojo against Python, Rust, and C++ on actual ML inference pipelines. MLIR compilation, ownership model, SIMD types, and why the '68,000x faster than Python' claim needs serious context.
mojo · python · mlir · simd · performance · machine-learning · systems-programming · rust · cpp · benchmarksARIA Live Regions, Screen Reader Internals, and Systematic Audit Methodology
A technical guide to web accessibility covering ARIA live regions, screen reader DOM parsing, focus management patterns, and systematic audit methodology with automated axe-core testing.
accessibility · aria · screen-reader · a11y · axe · focus-management · wcagScaling WebRTC to 10,000 Participants: mediasoup, LiveKit, and Janus Compared
A battle-tested comparison of WebRTC media servers. SFU vs MCU architecture, simulcast configuration, VP9/AV1 codec negotiation, and the scaling patterns that actually work when you need 10,000 concurrent participants.
webrtc · mediasoup · livekit · janus · sfu · mcu · simulcast · av1 · vp9 · real-time · video-streamingKubernetes Gateway API: Replacing Ingress with Envoy Gateway and Cilium
An analysis of migrating 200+ services from nginx-ingress to Gateway API with Envoy Gateway and Cilium mesh. HTTPRoute, TLS passthrough, traffic splitting, and why the Ingress resource was always a limited abstraction.
kubernetes · gateway-api · envoy · cilium · ingress · service-mesh · httproute · tls · traffic-splitting · infrastructureZig's Comptime: A Unified Approach to Compile-Time Systems Programming
Compile-time code execution in Zig replaces macros, generics, and conditional compilation in one unified mechanism. Rebuilding a packet parser in Zig reveals how comptime challenges assumptions embedded in traditional C project maintenance.
zig · comptime · systems-programming · c-interop · memory-safety · compile-time · low-level · infrastructureValkey vs Redis: Migrating 14TB of Production Data, What the Documentation Doesn't Cover
An analysis of Valkey's multi-threaded I/O, RESP3 protocol, and memory optimizations. Includes a real migration from Redis 7.2, benchmark methodology, and the gotchas encountered in production.
valkey · redis · in-memory-database · resp3 · performance · migration · multi-threaded-io · cachingSignals and Fine-Grained Reactivity: Angular, SolidJS, Preact, and the Shift from Virtual DOM Diffing
Fine-grained reactivity is winning. I traced the signal pattern from SolidJS through Angular 17+ and Preact Signals, dug into the TC39 proposal, and benchmarked why virtual DOM diffing is becoming an expensive relic.
signals · reactivity · angular · solidjs · preact · tc39 · virtual-dom · fine-grained-reactivity · frontend-architectureLoad Balancing Algorithms: Maglev Hashing, Consistent Hashing with Bounded Loads, and P2C
An analysis of modern load balancing algorithms, Google's Maglev hashing, consistent hashing with bounded loads, power of two choices, and production HAProxy/Envoy configurations.
load-balancing · maglev · consistent-hashing · p2c · haproxy · envoy · infrastructureWireGuard Mesh Networking: Tailscale and Headscale Internals Explained
How Tailscale builds a mesh network on top of WireGuard. DERP relay servers, NAT traversal, MagicDNS, and a hands-on guide to building your own private network overlay with Headscale.
wireguard · tailscale · headscale · mesh-networking · nat-traversal · derp · vpn · networkingLLVM MLIR: Building Custom Compiler Dialects for Domain-Specific Optimization
An examination of MLIR's multi-level intermediate representation architecture, custom dialect construction, and the convergence of ML compilers, hardware accelerators, and DSLs on the MLIR framework.
llvm · mlir · compilers · dialects · optimization · machine-learning · dsl · intermediate-representationHSMs and Cloud KMS: PKCS#11, Envelope Encryption, and Real Secrets Management Architecture
A deep technical walkthrough of hardware security modules, cloud KMS services, PKCS#11 interfaces, envelope encryption patterns, and building a secrets management architecture that actually survives audit.
hsm · kms · pkcs11 · envelope-encryption · key-rotation · aws-cloudhsm · secrets-managementCSS Anchor Positioning and Popover API: Implementing Tooltips Without JavaScript
Build tooltips, dropdowns, and floating UI entirely in CSS with the anchor() function, position-try-fallbacks, and the Popover API. An examination of current browser support and remaining specification gaps.
css · anchor-positioning · popover-api · frontend · tooltips · dropdowns · browser-apisKafka vs Redpanda: Event Streaming Benchmarks and Migration Analysis
A head-to-head comparison between Apache Kafka and Redpanda. Raft vs ZooKeeper, C++ vs JVM overhead, tail latency at p99, schema registry compatibility, and a detailed analysis of migrating 2TB/day of event streams.
kafka · redpanda · event-streaming · benchmarks · infrastructure · raft · zookeeper · p99-latencyWASM GC: Garbage-Collected Reference Types in WebAssembly
The WASM GC proposal brings struct and array types with automatic garbage collection to WebAssembly. This article examines its impact on Kotlin/WASM, Dart/WASM, and browser runtime architecture.
webassembly · wasm-gc · kotlin · dart · browser-runtimes · garbage-collection · performanceAI Orchestration Frameworks: Semantic Kernel vs LangChain vs LlamaIndex
A real comparison of AI orchestration frameworks, plugin architectures, memory management, and building a RAG pipeline with each to see where they actually differ when the rubber meets the road.
semantic-kernel · langchain · llamaindex · rag · ai-orchestration · llm · pluginsLLM Function Calling Patterns for Reliable Tool Use in AI Agents
A technical guide to LLM tool use, structured output, parallel tool calls, retry strategies, error recovery, orchestration patterns, and lessons from production AI agent development.
llm · function-calling · tool-use · ai-agents · structured-output · openai · anthropic · langchainAutomated Code Review: Semgrep Custom Rules, AST Pattern Matching, and Building Org-Specific Security Scanners
A practical guide to automated code review with Semgrep, writing custom rules with AST pattern matching, taint analysis for tracking untrusted data, building organization-specific security scanners, and integrating everything into CI for automated enforcement.
semgrep · ast · code-review · security · static-analysis · taint-analysis · custom-rules · ci-cdPostgreSQL CDC with Debezium: Logical Replication, WAL Decoding, and Real Event-Driven Architecture
A deep technical guide to PostgreSQL Change Data Capture with Debezium, logical replication slots, WAL decoding internals, the transactional outbox pattern, and building production event-driven architecture with Kafka Connect. With real configurations and gotchas from production.
postgresql · cdc · debezium · kafka · logical-replication · wal · event-driven · outbox-patternLinux io_uring: Async I/O Architecture and Node.js Integration
A systems-level analysis of io_uring, submission/completion queues, registered buffers, fixed files, kernel-side polling, and benchmarks comparing io_uring against epoll and synchronous I/O for various workloads.
io_uring · linux · async-io · nodejs · systems-programming · kernel · performance · epollTime Series Databases: TimescaleDB Hypertables, QuestDB Columnar Ingestion, and InfluxDB IOx
Building a real IoT data pipeline that handles 1M inserts/sec, comparing TimescaleDB hypertables, QuestDB's columnar engine, and InfluxDB IOx with actual benchmarks and production gotchas.
time-series · timescaledb · questdb · influxdb · iot · data-pipeline · performanceDNS Privacy Stack: DoH, DoT, DNSSEC, Encrypted Client Hello, and Comparative Privacy Analysis
A comprehensive guide to the DNS privacy stack, DNS-over-HTTPS, DNS-over-TLS, DNSSEC validation, Encrypted Client Hello (ECH), Oblivious DoH, with honest analysis of what each protocol actually protects against and a practical implementation guide.
dns · privacy · doh · dot · dnssec · ech · encrypted-client-hello · networking · securitygRPC + Protobuf + ConnectRPC: Type-Safe Streaming APIs with Code Generation
A practical guide to gRPC, Protocol Buffers, and ConnectRPC for building type-safe streaming APIs, with comparisons to REST and tRPC, reflection setup, load balancing considerations, and production-ready code.
grpc · protobuf · connectrpc · api-design · streaming · rpc · typescript · golangApache Arrow DataFusion: Building Custom Query Engines in Rust
A hands-on guide to Apache Arrow DataFusion, building custom query engines in Rust with columnar execution, push-based vs pull-based models, custom table providers, user-defined functions, and real-world performance tuning for analytical workloads.
arrow · datafusion · rust · query-engine · columnar · data-engineering · analytics · performanceStreaming HTML: Out-of-Order Streaming, Island Architecture, and Selective Hydration
An analysis of streaming HTML strategies across Marko, Qwik, and Astro, out-of-order streaming, island architecture internals, and why progressive hydration changes everything about web performance.
streaming-html · island-architecture · progressive-hydration · marko · qwik · astro · web-performanceReact Server Components: Streaming Architecture, Performance Tradeoffs, and Migration Patterns
A deep dive into React Server Components, streaming SSR, Suspense boundaries, and server actions, with honest benchmarks on the real cost of client components and practical migration patterns.
react · rsc · server-components · streaming · ssr · suspense · next.js · frontend-architectureSupply Chain Security: Sigstore, SLSA, SBOMs, and Securing the Build Pipeline End-to-End
A technical guide to software supply chain security, from Sigstore cosign keyless signing and SLSA framework compliance to SBOM generation with CycloneDX, npm provenance, and building a defensible pipeline that catches the attacks that matter.
supply-chain · sigstore · cosign · slsa · sbom · npm-provenance · security · ci-cd · build-pipelinePasskeys and WebAuthn: A Production Implementation Guide for FIDO2 Passwordless Authentication
A brutally honest walkthrough of implementing FIDO2 passkeys with WebAuthn, resident credentials, cross-device auth, attestation ceremonies, and all the edge cases that will ruin your weekend.
passkeys · webauthn · fido2 · passwordless · authentication · security · cryptographyPlatform Engineering: Building Internal Developer Platforms with Backstage and Port
A hands-on guide to building IDPs that developers actually want to use, covering Backstage plugins, Port blueprints, golden paths, and the developer experience metrics that matter.
platform-engineering · backstage · port · idp · developer-experience · golden-paths · self-serviceGPU Rendering Pipeline: Blend Modes, Porter-Duff Compositing, and Tile-Based Rendering
A deep dive into the GPU rendering pipeline, from blend modes and Porter-Duff compositing to tile-based rendering, Skia's GPU backend, Vello's compute-shader approach, and practical GPU profiling techniques for web content.
gpu · rendering · blend-modes · compositing · skia · vello · tile-based-rendering · browser · graphics-pipelineTypeScript 5.5: Inferred Type Predicates, Isolated Declarations, and Real-World Gotchas
Analysis of TypeScript 5.5's biggest features: inferred type predicates that actually work, isolated declarations for faster builds, config extends improvements, regex type narrowing, and patterns that will bite you.
typescript · type-predicates · isolated-declarations · type-narrowing · javascript · dx · compiler · type-systemLLM Inference Optimization: KV-Cache, PagedAttention, and GPU Memory Efficiency
An analysis of LLM inference optimization mechanics, from KV-cache management and PagedAttention in vLLM to continuous batching, speculative decoding, and FlashAttention-3, with throughput measurements from production deployments.
llm · inference · kv-cache · vllm · paged-attention · flash-attention · speculative-decoding · gpu · optimizationImmutable Infrastructure: Packer Golden Images, AMI Pipelines, Blue-Green Deployments, and the Real Migration from Mutable Servers
A practical guide to immutable infrastructure, building golden images with Packer, automating AMI pipelines, implementing blue-green deployments, eliminating configuration drift, and the honest story of migrating a production system from mutable servers.
immutable-infrastructure · packer · golden-images · ami · blue-green-deployments · configuration-drift · devops · infrastructure-as-code · deployment · automationContainer Runtime Internals: runc, youki, kata-containers, and Sandbox Isolation Models
Deep dive into container runtime internals, runc vs youki (Rust), kata-containers microVMs, gVisor sandboxing, cgroup v2 resource control, and namespace isolation with real syscall traces.
containers · runc · youki · kata-containers · gvisor · cgroups · namespaces · oci · kubernetes · securityNext.js 15 PPR: Partial Prerendering, Static Shells, and the End of the ISR vs SSR Debate
How Next.js 15's Partial Prerendering combines static and dynamic rendering in a single request. Static shell + dynamic holes via Suspense, real TTFB measurements, and a practical migration guide from ISR.
nextjs · ppr · partial-prerendering · react · suspense · performance · ssr · isr · static-generationCRDTs for Local-First Software: Building Collaborative Apps Without Servers
Deep dive into CRDTs with Yjs and Automerge 2.0. Operation-based vs state-based approaches, conflict resolution strategies, and real patterns for building collaborative applications that work offline.
crdts · local-first · yjs · automerge · distributed-systems · collaboration · offline-first · conflict-resolution · real-timeWebNN API: Hardware-Accelerated ML in the Browser, NPU Access, and Real On-Device Inference Benchmarks
A hands-on exploration of the WebNN API, accessing NPUs and GPUs from the browser for ML inference, ONNX model execution, real benchmark comparisons with WASM and WebGPU, and privacy-preserving AI patterns that keep data on the device.
webnn · machine-learning · browser-ml · npu · onnx · on-device-inference · webgpu · privacy · edge-ai · hardware-accelerationColumn-Oriented Storage Formats: Parquet, Lance, Delta Lake, and Iceberg Architecture and Optimization
Understanding column-oriented storage formats from the inside: row groups, predicate pushdown, dictionary encoding, and practical query optimization across Apache Parquet, Lance, Delta Lake, and Apache Iceberg.
parquet · lance · delta-lake · iceberg · column-storage · data-engineering · analytics · query-optimizationOpenTelemetry in Production: Distributed Tracing Implementation and Cost Control
Auto-instrumentation gotchas, custom spans that actually help, baggage propagation patterns, Jaeger vs Tempo showdown, trace-based testing, and how to not go bankrupt on observability costs.
opentelemetry · distributed-tracing · observability · jaeger · tempo · microservices · otel · spans · instrumentationUnikernels vs MicroVMs: Firecracker Internals, Unikraft, and the Real Cost of Serverless Isolation
A comprehensive comparison of Firecracker MicroVMs and Unikraft unikernels: boot time benchmarks, memory footprint analysis, KVM internals, and what actually provides isolation for serverless functions in production.
unikernels · firecracker · microvm · serverless · unikraft · kvm · virtualization · cloud-infrastructureTCP vs QUIC vs HTTP/3: Kernel Bypass, Zero-Copy, and Latency Optimization
Real benchmarks comparing TCP, QUIC, and HTTP/3 with io_uring, DPDK, and zero-copy networking. Congestion control algorithms dissected with actual latency measurements from production.
tcp · quic · http3 · io_uring · dpdk · zero-copy · kernel-bypass · networking · performance · congestion-controlCell-Based Architecture: Blast Radius Isolation, Shuffle Sharding, and How AWS and Slack Reduce Failure Domains
A deep technical look at cell-based architecture, how to isolate failures to a fraction of users through independent cells, shuffle sharding, cell routing, and real implementation patterns drawn from AWS, Slack, and production systems.
cell-based-architecture · blast-radius · shuffle-sharding · fault-isolation · distributed-systems · aws · slack · resilience · high-availability · architectureScroll-Driven Animations with CSS scroll-timeline, ViewTimeline, and the Web Animations API
Building performant scroll-driven animations without JavaScript using CSS scroll-timeline, animation-range, and the Web Animations API. Production examples replacing [GSAP](https://gsap.com/) ScrollTrigger and Intersection Observer.
css · scroll-animations · web-animations-api · scroll-timeline · viewtimeline · frontend-performance · css-animationsVector Databases Compared: Running the Same RAG Pipeline on pgvector, Qdrant, Pinecone, and Weaviate
Honest benchmarks comparing pgvector 0.8, Qdrant, Pinecone, and Weaviate for a real RAG pipeline. HNSW vs IVF indexing, quantization tradeoffs, latency at scale, and practical recommendations for different use cases.
vector-databases · pgvector · qdrant · pinecone · weaviate · rag · embeddings · hnsw · aiMLOps Feature Stores: Feast, Tecton, Real-Time Features, and Production ML Pipeline Architecture
A technical guide to feature stores in production ML, covering Feast and Tecton, point-in-time joins, real-time feature serving, feature drift detection, and the architecture that makes it all work without collapsing under its own complexity.
mlops · feature-stores · feast · tecton · machine-learning · real-time-features · point-in-time-joins · feature-drift · ml-pipelines · production-mlDeterministic Simulation Testing: Fault Injection Methodology in TigerBeetle and FoundationDB
Seed-based fault injection, deterministic scheduling, and the methodology behind finding impossible bugs in distributed systems. Real techniques from TigerBeetle and FoundationDB's testing frameworks.
deterministic-simulation · testing · distributed-systems · tigerbeetle · foundationdb · fault-injection · correctness · zigEffect-TS: Algebraic Effects and Typed Error Handling in TypeScript
An examination of Effect-TS: algebraic effects for TypeScript, structured concurrency, typed errors, Fiber-based scheduling, and why replacing try-catch chaos with a real effect system makes code dramatically more reliable.
effect-ts · typescript · algebraic-effects · error-handling · structured-concurrency · functional-programmingCapability-Based Security with WASI Preview 2: Least-Privilege Plugin Systems at the Runtime Level
How WASI Preview 2's capability-based model enforces least-privilege at the runtime level. Filesystem sandboxing, network capabilities, wasi-http, and production-grade secure plugin architectures.
wasi · webassembly · capability-based-security · wasm · sandboxing · plugin-systems · wasi-http · component-modelGraph Databases: Neo4j vs SurrealDB vs Apache AGE, Cypher, GQL, and Real Social Network Query Optimization
A hands-on comparison of three graph database approaches, Neo4j (native graph), SurrealDB (multi-model), and Apache AGE (PostgreSQL extension). Covers Cypher vs emerging GQL standard, property graph models, and real query optimization from a social network project.
graph-databases · neo4j · surrealdb · apache-age · cypher · gql · property-graphs · query-optimization · social-network · databasesGraviton4 ARM64 Migration Benchmarks: Real Production Performance Data
Production benchmarks from migrating workloads to AWS Graviton4 ARM64 instances. Node.js, Go, and Rust performance differences, NEON SIMD gotchas, compiler flags that matter, and actual cost savings after 4 months.
arm64 · graviton4 · aws · benchmarks · performance · migration · nodejs · go · rustScyllaDB: Cassandra-Compatible Low-Latency Database Architecture and Migration
A deep dive into ScyllaDB's shard-per-core architecture, Seastar framework internals, and real p99 latency benchmarks against Apache Cassandra. Plus a battle-tested migration guide.
scylladb · cassandra · distributed-databases · low-latency · seastar · cpp · nosql · shard-per-coreNix Flakes as a Docker Replacement for Local Development Environments
A practical guide to Nix Flakes for reproducible dev environments: flake.nix anatomy, devShells, replacing Docker for local development, cross-compilation, and a documented project migration with all the pain points.
nix · flakes · devops · docker · reproducible-builds · developer-experience · cross-compilationCrossplane: Infrastructure Composition on Kubernetes, Replacing Terraform with GitOps-Native IaC
A practitioner's guide to Crossplane, infrastructure composition on Kubernetes using XRDs, compositions, and provider-aws. How we replaced Terraform with a GitOps-native IaC approach and what we learned in the process.
crossplane · kubernetes · infrastructure-as-code · gitops · terraform · xrd · compositions · provider-awsAsync Rust Internals: Tokio Runtime Design, Work-Stealing, and Building a Minimal Runtime from Scratch
A deep technical exploration of how Tokio actually works, the work-stealing scheduler, I/O driver, timer wheel, waker mechanism, and the Pin story. Then we build a minimal async runtime from scratch to solidify the mental model.
rust · async · tokio · runtime · work-stealing · futures · waker · io-driver · concurrency · systems-programmingHTMX 2.0: Hypermedia-Driven Applications in Production
A technical analysis of HTMX 2.0's hypermedia-driven approach, hx-boost, WebSocket extensions, and server-sent events. Includes benchmarks against React and an assessment of when the approach is appropriate.
htmx · hypermedia · frontend · react · ssr · websockets · sse · spa-fatigueJIT Compilation Techniques: Cranelift, Copy-and-Patch, and CPython 3.13
An examination of modern JIT compilation techniques, Cranelift's approach to fast code generation, copy-and-patch compilation, tiered compilation strategies, register allocation, and the internal architecture of CPython 3.13's experimental JIT compiler.
jit · cranelift · copy-and-patch · cpython · compilers · register-allocation · tiered-compilation · wasmCSS :has() Selector: Quantity Queries, Parent Selection, and Previously Impossible Patterns
An examination of CSS :has(), from parent selection and sibling combinators to quantity queries and responsive patterns that genuinely could not exist before this selector shipped. Plus performance considerations that matter at scale.
css · has-selector · quantity-queries · responsive-design · frontend · performance · selectors · css-architectureDeno 2: Full Node Compatibility, Fresh 2.0, and Comparative Benchmarks Against Node 22 and Bun
An honest technical assessment of Deno 2's Node.js compatibility layer, npm specifiers, Fresh 2.0's island architecture, and performance benchmarks that tell a more nuanced story than the marketing suggests.
deno · node-js · bun · fresh-framework · javascript-runtimes · typescript · performanceSmithy: AWS's API Modeling Language and Building Your Own SDK Generator
An examination of Smithy, AWS's protocol-agnostic API modeling language, its code generation pipeline, custom validators, trait system, and how to build a custom SDK generator from a Smithy model.
smithy · api-design · aws · sdk-generation · code-generation · openapi · protocol-buffers · idlS3-Compatible Object Storage: MinIO, Tigris, R2, and Multipart Upload Internals
A technical walkthrough of S3-compatible object storage internals, from multipart upload mechanics and presigned URL security to building your own CDN layer with edge caching on top of MinIO, Tigris, and Cloudflare R2.
s3 · minio · tigris · cloudflare-r2 · object-storage · multipart-upload · presigned-urls · cdn · edge-caching · infrastructureZero-Knowledge Proofs for Authentication: Privacy-Preserving Identity Verification
How to implement ZK-based authentication using zk-SNARKs, Groth16, and Plonk. Circom circuit design, practical examples of proving identity without revealing credentials, and why ZK auth is useful far beyond Web3.
zero-knowledge-proofs · zk-snarks · groth16 · plonk · circom · authentication · cryptographyMicro-Frontends in 2026: Module Federation 2.0, Native Federation, and Import Maps
A comprehensive guide to micro-frontend architecture in 2026, Module Federation 2.0, Native Federation, import maps, shared state management, and a documented migration from a monolith SPA to independently deployable frontends.
micro-frontends · module-federation · native-federation · import-maps · webpack · rspack · shared-state · frontend-architectureWASI Threads: Shared Memory Parallelism Comes to WebAssembly
Shared memory WebAssembly, atomics, parallel computation, SharedArrayBuffer, exploring WASM threads through a ray tracer implementation to separate reality from hype.
wasi · wasm · threads · shared-memory · atomics · parallel-computing · webassembly · ray-tracerSQLite at the Edge: LiteFS, Turso, and Why Postgres Isn't Always the Answer
A practical guide to running SQLite as a distributed edge database. LiteFS replication internals, libSQL extensions, Turso's architecture, and real benchmarks showing when SQLite obliterates Postgres for read-heavy edge workloads.
sqlite · litefs · turso · libsql · edge-computing · distributed-databases · fly-ioConsensus Algorithms Compared: Raft, Multi-Paxos, Viewstamped Replication, and EPaxos
An implementation-focused comparison of distributed consensus algorithms, Raft, Multi-Paxos, Viewstamped Replication, and EPaxos, covering leader election, log compaction, liveness guarantees, and the nuances that only surface when you build them.
raft · paxos · epaxos · consensus · distributed-systems · viewstamped-replication · leader-election · log-replicationOpenAPI 4.0 (Moonwalk): The Spec Rewrite That Actually Matters
New overlay mechanism, first-class webhooks, reusable workflows, comparison with 3.1, and a real migration path from Swagger, OpenAPI 4.0 is the biggest API spec change in a decade.
openapi · moonwalk · api-design · api-specification · webhooks · swagger · rest-api · api-gatewayeBPF in Production: Building Kernel-Level Observability for Kubernetes Without the Overhead
How to use eBPF for production-grade observability in Kubernetes clusters. Cilium, Tetragon, custom kprobes, and why traditional monitoring is leaving performance on the table.
ebpf · kubernetes · cilium · tetragon · observability · linux-kernel · security-monitoringBrowser Storage APIs: OPFS, IndexedDB, and SQLite-over-WASM for Offline-First Applications
An analysis of modern browser storage APIs, Origin Private File System, IndexedDB transaction semantics, SQLite compiled to WASM, storage quotas, and offline-first architecture design.
opfs · indexeddb · sqlite · wasm · offline-first · storage-api · web-workers · browser-storageRate Limiting Algorithms: From Token Buckets to Distributed Lua Scripts
Token bucket, sliding window log, GCRA, distributed rate limiting with Redis, the algorithms behind every API gateway, with production-ready Lua script implementations.
rate-limiting · token-bucket · sliding-window · gcra · redis · lua · api-gateway · distributed-systemsV8 Maglev and Turboshaft: Mid-Tier and Backend Compiler Architecture
An analysis of V8's mid-tier JIT compiler Maglev and the Turboshaft IR that's quietly replacing Turbofan's Sea of Nodes. Real performance data, deoptimization mechanics, and why your 'optimized' code might be doing the opposite.
v8 · maglev · turboshaft · jit-compilation · javascript-performance · compiler-internalsMemory-Safe C++ Alternatives: Carbon, Cpp2, Val, and Circle Compared
A technical comparison of the emerging C++ successor languages, Carbon, Cpp2/cppfront, Val, and Circle, with real interop examples, migration cost analysis, and objective assessments of current maturity.
c++ · carbon · cpp2 · val · circle · memory-safety · systems-programming · cppfrontGraalVM Native Image: From 8-Second Spring Boot Startup to 50ms
AOT compilation for Spring Boot, reflection configuration challenges, memory footprint analysis, and production trade-offs from six months of operational experience.
graalvm · native-image · spring-boot · aot · jvm · startup-time · performance · javaPostgreSQL 17: Incremental Backup, Logical Replication Failover, and Operational Improvements
A practical deep dive into PostgreSQL 17's most impactful features: incremental backup with pg_basebackup, logical replication failover, new JSON functions, pg_stat_io insights, and real migration strategies from PG 15/16.
postgresql · databases · backup · replication · migration · pg17Web Components in 2026: Declarative Shadow DOM, CSS Parts, and Framework Interoperability
Declarative Shadow DOM, CSS parts, Lit 4.0, and universal custom element support. An assessment of Web Component maturity in 2026, covering SSR, framework interop, and remaining friction points.
web-components · shadow-dom · custom-elements · lit · css-parts · declarative-shadow-dom · frontend · interopLLM Quantization Compared: GPTQ, AWQ, GGUF, and SqueezeLLM
A practical comparison of LLM quantization methods with real perplexity benchmarks, memory footprints, and inference speed. When to use each method, and the surprising cases where 4-bit beats 8-bit.
llm · quantization · gptq · awq · gguf · inference-optimizationChaos Engineering in Production: Steady-State Hypothesis, LitmusChaos, and Blast Radius Control
LitmusChaos, Chaos Monkey, blast radius control, and the steady-state hypothesis, a technical guide to breaking things on purpose in controlled, productive ways.
chaos-engineering · litmus-chaos · chaos-monkey · gameday · sre · reliability · kubernetes · blast-radiusRust + WASM Component Model in Production: Polyglot Microservices That Actually Work
Building production polyglot microservices with Rust, the WASM Component Model, and WIT interfaces. Cross-language composition, wasm-tools workflows, and hard lessons from running this at scale.
rust · webassembly · component-model · wit · wasm-tools · microservicesIncremental Computation with Salsa and Adapton Frameworks
An examination of demand-driven incremental computation as implemented in rust-analyzer, memoization with automatic invalidation, dependency tracking, and the architectural patterns that enable sub-millisecond IDE responses.
incremental-computation · salsa · adapton · rust-analyzer · compilers · ide · demand-driven · memoizationBun v2 Runtime Internals: JavaScriptCore, Zig Allocators, and io_uring Syscall Batching
An analysis of Bun v2's architecture: JavaScriptCore vs V8 tradeoffs, Zig's arena allocator design, syscall batching with io_uring, and the engineering decisions behind the 3x cold start improvements.
bun · javascript · runtime-internals · zig · io_uring · performanceServerless Cold Starts: A Comprehensive Optimization Playbook
Provisioned concurrency, SnapStart, LLRT, V8 snapshots, a thorough benchmark comparison revealing which cold start optimization strategies actually deliver measurable results.
serverless · cold-starts · aws-lambda · provisioned-concurrency · snapstart · llrt · v8-snapshots · performanceWebGPU Compute Shaders for ML Inference: Client-Side Model Execution
Running ML models directly in the browser with WebGPU compute shaders. WGSL implementation details, benchmarks against WebGL and server-side inference, and practical limitations of client-side model execution.
webgpu · wgsl · machine-learning · browser-inference · compute-shaders