Hello! I'm Nikolay Donets

This blog explores the messy middle between AI possible and AI production-ready systems:

  • About transforming ideas into working solutions
  • Where documentation matters as much as algorithms
  • And solid processes enable innovations

Practical insights for builders who care about reliable performance and systems without crumbling.

Posts

Latest Top
3 minutes read

Self-Generated In-Context Learning Explained

How to boost language models without the need for massive training data

10 minutes read

Attacks on Large Language Models and LLM-based Services

Holistic approach needed to secure LLMs, balancing trade-offs and evolving threats. Ethical responsibilities and proactive measures essential.

5 minutes read

GEMBA for Machine Translation tasks

Review of the GEMBA approach for assessing translation quality using LLMs.

3 minutes read

Review of Guidance – Generative AI Framework for HMG

Review of the document about using generative AI safely and securely for civil servants and government workers.

19 minutes read

What is an Information Retrieval System?

A beginner-friendly introduction to Information Retrieval systems, this article highlights their key features and how they work.

14 minutes read

Prompt design for Large Language Models

The way the prompt is written has a significant impact on the LLMs output and the final task results.

3 minutes read

Large Language Models in Regulated Industries

Notes from my talk at the Large Language Models meetup in London on 13 June 2023.

7 minutes read

Making Code Efficient with LLMs. Part 2

Refactoring of pyEntropy library using LLMs Bard and Chat GPT. Part 2.

6 minutes read

Making Code Efficient with LLMs. Part 1

Refactoring of pyEntropy library using LLMs Bard and Chat GPT. Part 1.

6 minutes read

ABC and XYZ Analysis for Product Development

The ABC method is based on the Pareto principle. Use the XYZ method to categorize opportunities.

2 minutes read

Are facial recognition systems so bad?

London Police Facial Recognition system fails in 80% of cases. Is everything so bad?

9 minutes read

Mindmap on A/B tests designing and experiments

A summary from Designing and Experiment course to better understand the mechanics of tests.

See all

Apps

Calculator 42

4.6 stars from over 2200 happy users

Calculator 42 with History PRO is an iOS calculator with a simple and intuitive interface.

Projects

pyEntropy

pyEntropy GitHub Repo stars pyEntropy PyPI - Downloads

pyEntropy is a Python library for calculating entropy and mutual information.