Solving the ‘Blank Stare’ Problem: Integrating BM25 with RAG for Robust Information Retrieval

Susiloharjo

Hybrid Search BM25 RAG: Bridging Keyword and Vector Search In the landscape of modern retrieval systems, the “blank stare” problem remains one of the most significant hurdles for search architects. When a user queries a system for a specific technical concept rather than a literal keyword, traditional methods often fail. This is precisely where Hybrid … Read more

The Architecture of Trust: Decoding Indonesia’s New AI Content Labeling Mandate and the Future of Digital Provenance

Susiloharjo

The Architecture of Trust: Decoding Indonesia’s New AI Content Labeling Mandate and the Future of Digital Provenance In an era where synthetic media can mirror reality with uncanny precision, Indonesia’s recent move to implement a formal AI Content Labeling Mandate marks a significant turning point in the nation’s digital policy. As generative models proliferate, the … Read more

The Four-Agent Architecture: Decoding xAI’s Grok 4.20 and the Shift Toward Multi-Agent Problem Solving

Susiloharjo

The Four-Agent Architecture: Decoding xAI’s Grok 4.20 and the Shift Toward Multi-Agent Problem Solving The Four-Agent Architecture: Decoding xAI’s Grok 4.20 and the Shift Toward Multi-Agent Problem Solving When xAI released Grok 4.20 in March 2026, the announcement carried more weight than a typical version bump. Buried beneath the performance benchmarks and capability claims was … Read more

Space Computing: How NVIDIA’s Vera Rubin is Powering the Next Generation of Orbital AI Data Centers

Susiloharjo

Space Computing AI: How NVIDIA’s Vera Rubin Module is Building Orbital AI Data Centers Meta description: Space Computing AI reaches orbit with NVIDIA’s Space-1 Vera Rubin Module, delivering real-time autonomous control for lunar rovers and planetary infrastructure. The boundary between terrestrial data centers and deep-space computing infrastructure is collapsing. NVIDIA’s March 2026 announcement of the … Read more

The Vera Rubin Architecture: How NVIDIA’s H300 is Solving the Trillion-Parameter Efficiency Bottleneck

Susiloharjo

The Vera Rubin Architecture: NVIDIA’s Essential Answer to the Trillion-Parameter Efficiency Bottleneck The NVIDIA Vera Rubin platform redefines trillion-parameter AI training with a 10x cost reduction, unified HBM4/5 memory, NVLink 5, and a dedicated Physical AI foundry. The Scale Problem Nobody Talks About Training a language model with a trillion parameters is not a software … Read more

Beyond Chatbots: How ‘Stripe Minions’ and Agentic AI are Redefining Organizational Structures

Susiloharjo

Beyond Chatbots: How Agentic AI is Redefining Organizational Structures from Heroism to Stewardship

The software industry stands at a inflection point. For decades, engineering organizations have operated on a model of human heroism: skilled engineers diagnose production fires, manually patch critical systems, and carry institutional knowledge that no ticket tracker can capture. That model is now being stress-tested at scale—and a growing number of engineering leaders are asking whether heroism is a feature or a bug.

At Stripe, the question yielded an unexpected answer. The company’s internal agentic infrastructure—informally called the “Minions” system—now generates and merges approximately 1,300 pull requests per week with no direct human coding. The system’s architecture offers a rare, detailed look at how autonomous AI agents can operate within a production engineering environment, and more importantly, what their existence means for how organizations are structured.

Read more