CACogniArc

Personal Research Blog

CogniArc

Research notes on AI agents and engineering systems

A bilingual journal for research ideas, implementation patterns, and practical observations from building AI-native software.

Research interests

AI agentsSoftware engineeringDeveloper tooling

Recent blog posts

View all posts

Mar 8, 2026 | 1 min

Building CogniArc as a Research Journal

Why this site exists, what it tracks, and how the writing process stays tied to implementation.

Read ->

Mar 4, 2026 | 1 min

Notes on Agent Memory Retrieval

A practical comparison between context windows, vector search, and hybrid memory pipelines.

Read ->

Feb 26, 2026 | 1 min

Observability for Agent Tooling

Instrumentation patterns that make agent failures debuggable in production-like workflows.

Read ->

Featured research projects

View all projects

Agent Evaluation Workbench

Benchmarking multi-step tool-use reliability with reproducible test suites.

Next.jsTypeScriptPostgreSQL

Memory Retrieval Study

Comparing context window, vector retrieval, and hybrid memory pipelines.

PythonFAISSFastAPI

Developer Tooling Observability

Tracing task latency and failure modes across agent orchestration layers.

OpenTelemetryNode.jsGrafana

Need a quick guide?

Use the built-in assistant to find relevant sections and reading paths.

Open assistant