Skip to content
View Panolof's full-sized avatar

Block or report Panolof

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Panolof/README.md

Pano Pouroullis

AI Data Scientist building safety infrastructure for AI-assisted workflows.

What I work on:

  • Evaluation pipelines for frontier model behaviour (Inspect AI, multi-turn, cross-condition)
  • Cross-model verification — different model families catching what same-family review misses
  • Agentic safeguards — permission architectures, deterministic scripts, and hook systems that constrain AI agent behaviour
  • Mechanistic interpretability (circuit tracing, sparse autoencoders)

Writing: observatoryledger.com

BEng Electronic Engineering (Cum Laude), University of Pretoria. Python, TypeScript, SQL.

Pinned Loading

  1. inspect-ai-sycophancy-eval inspect-ai-sycophancy-eval Public

    Does a model keep the right answer when a user pushes back? A deterministic Inspect AI eval, with one dated two-model run.

    Python

  2. rag-vector-search rag-vector-search Public

    Local-first RAG demo with Milvus Lite, deterministic embeddings, relevance gating, a Flask UI, and an optional OpenAI CLI adapter.

    Python 2