alper nebi kanlı
writingprojectsabout

alper nebi kanlı

Staff AI / Machine Learning Engineer

Staff AI engineer with ten years of building production machine learning systems, from ranking models for search serving millions of users to LLM-based retrieval and agents, and a strong interest in how models work inside. Currently building hybrid retrieval, RAG and agentic search at Optimizely. Writes and speaks about embeddings, retrieval and interpretability, including Retrieval Explorables and talks at PyData London and AI Signals.

London, United Kingdom[email protected]www.alpernebikanli.comwww.linkedin.com/in/alpernkanli
Alper Nebi Kanlı in Bali

Experience

Optimizely

Staff Software Engineer - AI SearchJan 2026 - Present
Content Graph TeamLondon, England, United Kingdom
  • Continuously improving the hybrid search and ranking capabilities of the Content Graph project.
  • Building RAG and agentic search capabilities, including a hybrid RAG system that powers dynamic experiences on customer pages.
  • Leading the team's implementation of LLM-based workflows for task assessment and PR review.
  • Mentoring an engineer through the formal mentoring programme, guiding their study of LLM engineering.

Searchlight Cyber

AI EngineerJul 2025 - Nov 2025
AI TeamUnited Kingdom
  • Developed agentic AI solutions for cybersecurity and threat intelligence.

Hepsiburada

Senior Data ScientistJun 2024 - Jul 2025
Search Engine Teamİstanbul, Türkiye
  • Developed and trained learning-to-rank models for product search serving millions of users.
  • Built multi-objective ranking by designing training labels that combine several user-behaviour signals and blending scores from multiple models at ranking time.
  • Increased conversion rate (CR) and click-through rate (CTR), validated through detailed A/B test analyses.
  • Developed an ontology extraction pipeline using LLMs, and improved search with NLP techniques, LLMs and embeddings.
  • Managed production fine-tuning pipelines for BERT-style models, orchestrated with Airflow.
  • Improved MLOps pipelines, experiment tracking processes and evaluation frameworks.

Turkcell

Senior Data ScientistApr 2023 – Jun 2024
Computer Vision Teamİstanbul, Türkiye
  • Built an image search service by combining computer vision techniques with NLP models.
  • Created high-performance computer vision services, including a NIST-evaluated face recognition service.
  • Developed near-real-time ML services for edge devices such as Nvidia Jetson and Qualcomm.
Senior Search/NLP EngineerDec 2019 – Apr 2023
Search Engine Teamİstanbul, Türkiye
  • Owned a full rebuild of the search system, including the technology-selection decision.
  • Developed ranking algorithms and trained ranking models, including multi-objective ranking, and used LLMs and embeddings to improve relevance for search serving millions of users.
  • Implemented search APIs handling high request volumes, and developed Elasticsearch plugins.
  • Built a question answering system using knowledge graphs and NLP models.
  • Created ETL pipelines for search data to support algorithm development.
  • Mentored team members, contributed to architectural decisions, and served as Scrum Master.

Azerion

Senior Data ScientistJul 2018 – Nov 2019
Lely Projectsİstanbul, Türkiye
  • Developed machine learning solutions for forecasting and classification, utilizing LSTMs, RNNs, and ensemble methods like XGBoost.
  • Designed and implemented data engineering solutions for large-scale data processing.

Cybersoft

Data ScientistDec 2017 – Jul 2018
R&D Departmentİstanbul, Türkiye
  • Implemented data preprocessing and machine learning algorithms for a distributed environment.
  • Developed automated machine learning model selection and hyperparameter optimization services.
  • Supervised and mentored three intern groups, co-authoring conference papers with them.

UniteBT

RPA DeveloperJun 2017 – Nov 2017
RPA Teamİstanbul, Türkiye
  • Delivered RPA solutions to optimize business processes for various clients in a fast-paced environment.

Bilkon Ltd

Machine Learning EngineerAug 2016 – May 2017
R&D DepartmentAnkara, Türkiye
  • Developed components of a video management system, including machine learning pipelines.

Writing, Talks & Community

  • Retrieval Explorables (retrieval-explorables.com): interactive learning pages on retrieval and interpretability, presented at PyData London in November 2026.
  • Technical blog (alpernebikanli.com/blog): long-form essays on embeddings, LLMs, retrieval and transformer models.
  • Talks at PyData London (embedding space, Dec 2025; AI agents as multipliers, Jan 2026; Retrieval Explorables lightning talk, Nov 2026) and AI Signals (how LLMs represent concepts internally, Sep 2026).
  • Co-organizer of AI Signals, a London AI meetup series.

Skills & Technologies

Languages: Python, Java, TypeScript, JavaScript, C++, C#, SQL, Cypher, MATLAB

LLMs & Retrieval: RAG, Agentic Systems, Claude API, Claude Code, Embeddings, Hybrid Search, Learning to Rank, OpenSearch, Elasticsearch, LangChain, LangGraph, Langfuse, Neo4j, ChromaDB, Pinecone

Machine Learning: PyTorch, TensorFlow, BERT-Style Transformers, Scikit-Learn, XGBoost, CatBoost, NumPy, Pandas

MLOps & Cloud: GCP, AWS, Azure, Cloudflare, SageMaker, Docker, Kubernetes, Airflow, MLflow, Feast, GitLab CI/CD

Backend & Data: FastAPI, Flask, Django, Spring Boot, Node.js, PostgreSQL, MongoDB, Redis, Cassandra, BigQuery, Spark, Kafka, Flink, Hadoop, Hive

Education

Middle East Technical University2018

BSc, Electrical and Electronics Engineering

Udacity2020

Certification, Natural Language Processing Nanodegree

Papers & Patents

  • Implementation of Data Preprocessing Techniques on Distributed Big Data Platforms, 2019 4th International Conference on Computer Science and Engineering (UBMK)
  • Data Mining Library for Big Data Processing Platforms: A Case Study - Sparkling Water Platform, 2018 3rd International Conference on Computer Science and Engineering (UBMK)
  • Implementation of Association Rule Mining Algorithms on Distributed Data Processing Platforms, 2019 4th International Conference on Computer Science and Engineering (UBMK)
  • The Impact of RFF and Ensemble Modeling Approach in an E-commerce Product Ranking, 2025 33rd Signal Processing and Communications Applications Conference (SIU)

Interests

  • Cinema and Theatre
  • Reading widely (science, history, fiction, linguistics, philosophy)
  • Learning electric guitar
  • Sports
substacklinkedingithubblueskyrss

© 2026 Alper Nebi Kanlı