Marcio Cunha

Technical Articles

Practical articles, in-depth tutorials, and architectural insights on software engineering, AI, and systems.

All categoriesArquiteturaAutomaçãoBackendCarreiraDevOpsEletrônicaEngenhariaFrontendHardwareIANegóciosProdutividadeRedes
HardwareIA•4 min

Panther Lake: Unifying CPU, GPU, and NPU on a Single Silicon

Explore how Intel's Panther Lake architecture merges central processing units, integrated graphics, and artificial intelligence accelerators into a single silicon die, redefining energy efficiency and performance standards in personal computing.

IAHardware•4 min

Language Model Inference Optimization on Dedicated Hardware with Tensor Processing Unit Accelerators

Explore how Tensor Processing Unit accelerators optimize large-scale language model inference in high-demand production environments. We analyze hardware architectures, XLA compilation strategies, and real operational trade-offs.

IAArquitetura•5 min

Episodic and Procedural Memory Architecture in LLM-Based Autonomous Agents

Learn how to build robust autonomous agents in production using hybrid RAG, graph databases, and vector stores for long-term persistence and coordinated decision-making.

IABackend•4 min

Complex Document Information Extraction Pipeline Architecture with Vision and Language Models

Learn how to build resilient data flows to extract information from complex PDFs, tables, and receipts using visual artificial intelligence and language models.

IAEngenharia•5 min

Building Hallucination Evaluation Systems for LLM-Based AI Agents with Tool Use

Learn how to build reliability guardrails and evaluation systems for artificial intelligence agents that use external tools, reducing errors and fabricated responses in enterprise environments.

IABackend•4 min

Semantic Search Architecture with Qdrant and HNSW Indexing Under High Concurrency

Learn how to build a high-performance vector database infrastructure using Qdrant and the HNSW algorithm to handle millions of simultaneous queries in artificial intelligence systems.

IADevOps•7 min

Efficient Fine-Tuning of Large Language Models with LoRA and QLoRA on Local Infrastructure

Learn how to adapt large language models using low video memory on local servers by combining cost-effective fine-tuning techniques.

IAEngenharia•2 min

Building Evaluation Systems for Large Language Models with Semantic Unit Testing

Learn how to build semantic unit testing pipelines to guarantee the quality of LLM responses. Discover how to validate AI behaviors in an automated and deterministic manner.

IABackend•4 min

Implementation of Unstructured Data Extraction and Normalization Systems with Language Models and Distributed Pipelines

Learn how to design resilient distributed architectures and data pipelines combining artificial intelligence and batch processing to transform complex documents into structured information.

IABackend•4 min

Evaluation and Mitigation of Hallucination Risks in Retrieval-Augmented Generation Pipelines Using Critique Models

Learn how to combat hallucinations in Artificial Intelligence systems that query external databases, using critique models to validate the fidelity and truthfulness of generated answers in real time.