Are Agents Probabilistic Automation?
A small ambiguity in a coding agent session made me think about long-running uncertainty, explicit workflows, and why serious agent systems need guardrails.
Read MoreTechnical writing on ML, LLMs, and AI systems
A small ambiguity in a coding agent session made me think about long-running uncertainty, explicit workflows, and why serious agent systems need guardrails.
Read MoreThree AI agents used Agent Relay to review and improve Agent Relay itself. They found 7 issues, shipped SDK fixes, and rated the experience 9/10 -- with zero human involvement.
Read MoreI ran 200 benchmarks comparing two LLM tool-calling patterns -- standard ReAct loops vs generating complete Python scripts. The numbers surprised me.
Read MoreHere I am going to explain the windowing method of getting a cropped image with desired aspect ratio.
Read on MediumSetting up selenium on python to extract information from emails.
Read on MediumHand engineered filters and testing custom filters to extract features for applications.
Read on Medium