Ivan Zdravkov

Doctoral researcher in adaptive LLMs at the Institute of Robotics, Bulgarian Academy of Sciences. Also a backend engineer with 20+ years in production. MSc in AI & Data Processing.

Sofia, Bulgaria Joined on Oct 28, 2026 https://zdravkov.info

Ivan Zdravkov

Oct 28

Your model is not the product. The pipeline around it is.

#ai #machinelearning #programming #datascience

5 min read

Ivan Zdravkov

Oct 28

What the EU AI Act actually requires of a masking layer

#ai #privacy #security #research

6 min read

Ivan Zdravkov

Oct 28

What EWC buys you that flat L2 does not

#ai #machinelearning #research #deeplearning

5 min read

Ivan Zdravkov

Oct 28

What an internal MCP server actually solves

#ai #programming #opensource #llm

6 min read

Ivan Zdravkov

Oct 28

The real trade-off between an API model and a local one

#ai #llm #opensource #machinelearning

6 min read

Ivan Zdravkov

Oct 28

The demo is the easy part

#ai #programming #machinelearning #opensource

5 min read

Ivan Zdravkov

Oct 28

Permissions that travel with the person

#ai #security #programming #privacy

6 min read

Ivan Zdravkov

Oct 28

The masking layer matters more than the model

#ai #privacy #security #llm

7 min read

Ivan Zdravkov

Oct 28

LoRA on a 16 GB card: what fits and what does not

#ai #pytorch #deeplearning #opensource

5 min read

Ivan Zdravkov

Oct 28

Why I pre-register a kill-gate for every experiment

#ai #research #datascience #programming

5 min read

Ivan Zdravkov

Oct 28

How wrong is a model's confidence?

#ai #machinelearning #research #datascience

5 min read

Ivan Zdravkov

Oct 28

Fluency is not understanding

#ai #llm #nlp #research

5 min read

Ivan Zdravkov

Oct 28

Detecting PII before the model sees it

#ai #privacy #security #nlp

5 min read

Ivan Zdravkov

Oct 28

Catastrophic forgetting, measured

#ai #machinelearning #research #deeplearning

5 min read

Ivan Zdravkov

Oct 28

Blackwell quantisation on my own card

#ai #pytorch #deeplearning #research

5 min read

Ivan Zdravkov

Oct 28

An honest look at LLM-as-a-judge

#ai #llm #research #machinelearning

5 min read