March 2026

I realized one day that software alone became mechanical. I developed a taste for simplicity and problems started becoming less like learning experiences, and turned into mechanical exercises that resembled busy work more than true exploration. I realized I was working always in the abstract. I took detours, sure. I got multiple cyber security/pentesting certs, and sure it was interesting. But, it didn’t scratch that itch in the end. I realized I needed to start touching the physical […]

Ver mais

The internet just underwent its first brain surgery. No one announced it. No ribbon was cut. But inside the architecture of digital intelligence, something fundamental has moved, and most people haven’t noticed yet. Here’s what that shift looks like in practice. Watch How This Works Agent A: monitors real-time logistics data from shipping ports. It notices congestion emerging in Rotterdam Agent B: specializes in historical weather and delay patterns. Agent A wakes it up and hands off the […]

Ver mais

The increasing complexity of neural networks poses significant challenges for democratizing FL on resource?constrained client devices. Parallel split learning (PSL) has emerged as a promising solution by offloading substantial computing workload to a server via model partitioning, shrinking client-side computing load, and eliminating the client-side model aggregation for reduced communication and deployment costs. Since PSL is aggregation-free, it suffers from severe training divergence stemming from gradient directional inconsistency across clients. To address this challenge, we propose GAPSL, a […]

Ver mais

Crypto marketing got harder the moment attention became cheap and trust became expensive. That sounds dramatic, but the numbers explain why. Telegram said it crossed 1 billion monthly active users in March 2025, while Discord reports 200 million-plus monthly active users. Web3 projects now launch into platforms that are already saturated with communities, narratives, promos, and recycled hype. At the same time, a16z estimated there were roughly 30 to 60 million monthly active crypto users in 2025, which […]

Ver mais

TL;DR: Semantic RAG assumes the query embedding lands near the answer embedding. For multi-step questions — comparisons, computations, cross-document analysis — that assumption fails. Here are five architectural patterns that fix this: embrace agents over pipelines, separate storage by data type, route deterministic operations to deterministic tools, show your work, and build systems that know when they don’t know. Semantic RAG rests on a core assumption: the vector embedding of your query will be close to the vector […]

Ver mais

Decentralized Federated Learning (DFL) remains highly vulnerable to adaptive backdoor attacks designed to bypass traditional passive defense metrics. To address this limitation, we shift the defensive paradigm toward a novel active, interventional auditing framework. First, we establish a dynamical model to characterize the spatiotemporal diffusion of adversarial updates across complex graph topologies. Second, we introduce a suite of proactive auditing metrics, stochastic entropy anomaly, randomized smoothing Kullback-Leibler divergence, and activation kurtosis. These metrics utilize private probes to stress-test […]

Ver mais

(A senior engineer perspective on bucket metadata, negative caches, and surviving random-key floods) Introduction At internet scale, object storage is less about moving bytes and more about answering questions quickly: does this bucket exist, is this caller allowed, where is this object, and what exactly should we return? Those are metadata questions. This is why metadata becomes the real bottleneck. Every request touches it, including failures. A 404 still requires the system to prove absence. A HEAD is […]

Ver mais

Large Reasoning Models (LRMs) have shown exceptional reasoning capabilities, but they also suffer from the issue of overthinking, often generating excessively long and redundant answers. For problems that exceed the model’s capabilities, LRMs tend to exhibit the overconfidence phenomenon, generating overly short but incorrect answers, which may contribute to suboptimal performance. To address these issues, we propose Difficulty-Differentiated Policy Optimization (DDPO), an efficient reinforcement learning algorithm that optimizes simple and complex tasks separately based on the overconfidence phenomenon. […]

Ver mais

Large Reasoning Models (LRMs) have shown exceptional reasoning capabilities, but they also suffer from the issue of overthinking, often generating excessively long and redundant answers. For problems that exceed the model’s capabilities, LRMs tend to exhibit the overconfidence phenomenon, generating overly short but incorrect answers, which may contribute to suboptimal performance. To address these issues, we propose Difficulty-Differentiated Policy Optimization (DDPO), an efficient reinforcement learning algorithm that optimizes simple and complex tasks separately based on the overconfidence phenomenon. […]

Ver mais

The strong performance of large vision-language models (VLMs) trained with reinforcement learning (RL) has motivated similar approaches for fine-tuning vision-language-action (VLA) models in robotics. Many recent works fine-tune VLAs directly in the real world to avoid addressing the sim-to-real gap. While real-world RL circumvents sim-to-real issues, it inherently limits the generality of the resulting VLA, as scaling scene and object diversity in the physical world is prohibitively difficult. This leads to the paradoxical outcome of transforming a broadly […]

Ver mais