A bilingual personal research site built for steady writing, project case studies, and low-maintenance publishing.
Agent / LLM Research & Building
Hi, I’m sikm.
Agent / LLM Algorithm Graduate Student
This site is where I keep writing, project notes, research summaries, and a small amount of personal context around how I build AI systems.
Current Focus
Current focus
A few themes I am actively pushing forward.
LLM Agent Systems Multi-Agent Evaluation Tool Use Reliability RAG / MCP / Sandbox Object Detection Notes
Featured
Featured projects
Case studies that show problem framing, technical choices, and outcomes.
A staged agent workflow for log diagnosis, evaluated through real failure buckets and evidence quality.
Python Agent Workflow Prompting Evaluation
Latest
Latest writing
Ongoing notes on agent systems, evaluation, tooling, and implementation details.
May 10, 2026 · 7 min read
Rethinking Agent Design Through State
State transitions and explicit boundaries usually stabilize an agent system more than endless prompt tweaking.
Agent LangGraph Workflow
Latest
Research notes
Paper reading, experiment observations, and question breakdowns.
May 4, 2026 · 5 min read
RF-DETR Reading Note: Start from the Task, Not Only the Metric
A research note is stronger when it reconnects paper results with task framing, data conditions, and failure modes.
Research Object Detection RF-DETR
Latest
Life notes
A small amount of human context beyond pure technical output.
May 2, 2026 · 4 min read
Weekly Reset: Reordering Inputs, Projects, and Attention
Less a productivity trick, more a small rhythm that keeps long-term work from drifting.
Journal Reflection Workflow