saengil AI
  • journal
  • projects
  • about

projects

active research, experiments, and technical studies — frameworks for the curious mind.

active

RUNES protocol

a declarative framework for composable agentic behaviors — skill graphs, policy routing, and emergent task decomposition for persistent AI systems.

agentic frameworks python orchestration
active

ideonexus

an agentic research framework for managing AI/ML projects — tracking ideas with genealogy, designing sandboxed experiments, and monitoring the competitive landscape.

research ml knowledge graph
research

browser agent v2

next-generation autonomous browser automation with vision-language models, persistent session memory, and multi-step task planning.

automation vlm playwright
active

edge inference lab

benchmarking and optimizing quantized LLMs on constrained hardware — Jetson Orin, Raspberry Pi, and custom FPGA accelerators.

edge ai quantization hardware
research

semantic memory engine

a study of embedding-based memory consolidation — how agents can extract durable knowledge from ephemeral interactions and build coherent world models.

embeddings rag memory
archived

multi-model arena

a systematic evaluation framework for comparing LLM outputs across accuracy, coherence, creative reasoning, and cost efficiency dimensions.

evaluation benchmarks llm
saengil AI

the science of imagination — systems that think, create, and evolve.

explore

  • journal
  • projects
  • about

connect

  • github
  • contact
© saeng-il ai — YXZYS saeng-il ai [development]