Steven Lopez

I validated a Pulsar-to-Kafka bridge pipeline on StreamKernel this week — 253,235 records, zero loss, peak throughput of 15,647 records/sec on a laptop with co-located Docker brokers. github.com/IntuitiveDesig… #transport #streamkernel #pulsar #kafka

Most platforms solve pieces of the streaming AI pipeline. StreamKernel was built to collapse the stack: ✅ Transport ✅ Sinks ✅ In-process ONNX inference ✅ MLflow model governance ✅ Security + policy ✅ Air-gapped deployment ✅ JVM-native operations github.com/IntuitiveDesig…

Stop wasting time on network hops. 🚀 In-process AI enrichment means lower latency, better security for air-gapped apps, and a smoother write path. Same JVM. One process. github.com/IntuitiveDesig… #streamkernel #inprocess #ai #secure

Enterprise AI should not require runtime connectivity or remote inference. StreamKernel keeps inference, governance, security, and lineage inside one JVM pipeline. github.com/IntuitiveDesig… #AIInfrastructure #EnterpriseAI #AirGapped #StreamKernel

Everything runs in-process. No agents. No sidecars. StreamKernel embeds ONNX/DJL AI inference directly inside the JVM pipeline runtime — no external model server, no gRPC call, no network hop. github.com/IntuitiveDesig… #streamkernel #ai #onnx #streaming #pipeline #mlflow

Regulated environments reject phone-home architectures Defense, intelligence, healthcare, and critical infrastructure ops can’t afford a model that dials out to validate a license or fetch weights at runtime. github.com/IntuitiveDesig… #streamkernel #airgapped #banking #defense

Before Kafka. Before MongoDB. Before Spark, Before Pulsar, Before mTLS. Just the pipeline engine, wide open — 10 million events per second, zero drops, 3 billion records in five minutes. That's the ceiling. Everything else is your sink. github.com/IntuitiveDesig…

StreamKernel brings private AI, policy enforcement, transformation, and governed streaming delivery into a single bank-controlled JVM pipeline designed for restricted and air-gapped environments. github.com/IntuitiveDesig… #secure #banking #ai #airgapped #streamkernel #onnx

I’ve spent the last several months benchmarking a stream processing engine I’m building called StreamKernel. Not on a cloud cluster. Not on a tuned 96-core rack. github.com/IntuitiveDesig… #streamkernel #kafka #mongodb #airgapped

In the MongoDB WiredTiger benchmark article by Mark Callaghan, a server-class machine with 40 hyperthreads and 144 GB RAM reached 140,142 docs/sec on the larger-document insert test and 247,651 docs/sec on the small-document best case. #StreamKernel #MongoDB #WiredTiger

Streaming AI/Data pipeline kernel that runs source ingestion, policy enforcement, transformation, caching, sink delivery, DLQ routing, and metrics inside a single JVM. #AI #Java #PerformanceEngineering #StreamingData #Kafka #VectorSearch #EnterpriseArchitecture

Delta Lake + MLflow profile — in-process ONNX inference, model pulled from MLflow Registry, enriched records committed to Delta Lake with ACID guarantees. Verified by an independent Spark session No - sidecar embedding service 👇 @lopezstevie/stream-infer-persist-prove-0c3298b7e8cb" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/s… #MLOps #DeltaLake #MLflow

Every time your app ships data to a remote model server, you're paying a toll. In milliseconds. In dollars. 💵 In compliance risk. Most teams don't see it until it's load-bearing and too late. The AI Execution Tax — and how to eliminate it. @lopezstevie/the-ai-execution-tax-why-intelligence-must-move-closer-to-the-data-63e14c7c7e2c" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/t…

563M records. 955K ops/sec. One JAR. One broker. One laptop. 🔥 New benchmark — the Kafka producer ceiling, and what it costs to add transforms, guarantees, and security at scale. Full breakdown →@lopezstevie/streamkernel-the-kafka-ceiling-2c46a618502d" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/s… #StreamKernel #Kafka #DataEngineering #AI

The WiredTiger "write ceiling" is the next Network Tax. 📉 Stop fighting latch contention. Collapse the stack instead. How StreamKernel breaks it → @lopezstevie/streamkernel-the-mongodb-write-ceiling-4e4c2916c59c" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/s… #StreamKernel #MongoDB #WiredTiger #AI #Inference

The "Network Tax" is killing production AI. 📉 We must collapse the stack. If your data isn't where your inference happens, you're already behind. My latest on why AI is a data-management problem: @lopezstevie/the-future-of-ai-might-look-more-like-databases-than-chatbots-348766b3f564" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/t… #Inference #Kafka #StreamKernel #FrugalAI

Open source Enterprise RAG Connector Kit! - composable source adapters - batching, validation, retry, and structured logging - MCP tool - test coverage for core indexing and retrieval paths - and MORE! github.com/IntuitiveDesig…

We can't keep throwing more hardware at inefficient data pipelines. 🌱 I'm sharing my approach to Frugal AI Infrastructure—eliminating waste and collapsing silos with StreamKernel. Join the movement: @lopezstevie/streamkernel-f15e01d5e85f" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/s… #SustainableAI #GreenComputing #Technology

Stop paying the "Network Tax" for AI Inference. 🛑 I’m releasing the StreamKernel Exactly-Once Baseline. It’s time to collapse the infrastructure and move toward frugal, high-efficiency data streams. READ: @lopezstevie/streamkernel-kafka-exactly-once-baseline-c058f52b40f9" target="_blank" rel="nofollow noopener">medium.com/@lopezstevie/s… #Kafka #AI #Inference #DataEngineering