About
I am Alper Nebi Kanlı, a Staff Software Engineer at Optimizely in London. My full-time work is mostly search and information retrieval, OpenSearch and embeddings. More broadly I work across search engineering, natural language processing, and LLM systems.
This site is my long-form notebook for understanding the machinery underneath modern AI: embeddings, retrieval, language models, search systems, and the interfaces that make those systems useful.
I am most interested in AI as a way to extend how we read, search, reason, and create. The writing here is deliberately slower than a social post: I prefer to follow the mechanism, define the pieces, and make the system feel inspectable.
Outside the technical work, I care a lot about cinema, theatre, language and philosophy. Those interests tend to sneak into how I think about technology. On the side, I also help organise AI Signals meet-ups.
What I Write About
- Embeddings, latent spaces, and representation learning.
- LLMs, retrieval, agents, and the systems around them.
- Search architectures, especially Elasticsearch and Lucene.
- AI interfaces as tools for thought.
Talks
- AI Agents as Multipliers: What They Improve and What They Don't — PyData London · January 2026
- Understanding Embedding Space: Surprises, Pitfalls, Intuitions — PyData London · December 2025
- Geniş Dil Modelleri ve Uygulamaları (Large Language Models and Their Applications) — Yaşar University, İzmir · December 2024