Method

Synthetic data generation (LLM)

Generating training or evaluation data with LLMs (Evol-Instruct, self-instruct).

35 cases from 32 companies apply Synthetic data generation (LLM). 5 were checked in an online A/B test; 2 report results in production.

Typical tasks: Evaluate models, Build an assistant over internal docs, Rank search results, Label data with LLMs and 7 more. Companies: Pinterest, Airbnb, Zomato, Databricks, Avito, Celsus and 26 more. Often combined with: LLM-as-a-judge, Supervised fine-tuning, RAG, Hybrid search, LoRA fine-tuning and 85 more.

From Activity to Intent: Generating User Journeys with LLMssocial2026-10
CASTLE: training natural language search before the first real querytravel2026-10
Building Gavel: AI evaluation infrastructure across Eternalfood-delivery2026-10
Thinking Fast & Slow: How Databricks Built High-Speed and Deep Research Agentsgeneral2026-06
Doubling Medical Safety: Fine-Tuning Open LLMs for Women's Health Without Human Labelshealthtech2026-06
How a 4B Model Beat GPT at Agentic Retrieval (And How to Train One on a Neon Postgres Corpus)general2026-06
With Eyes Wide Shut: Detecting Sleeping People and Other Remote Verification Attacks Without Consentfintech2026-05
How We Do RL for Our LLM at Avitoclassifieds2026-05
Developing a RAG System on the History of Bashkortostan Using a Benchmarking Methodologygeneral2026-05
From Zero to Agents: Growing a Multi-Agent LLM Platform Inside a Companygeneral2026-05
Building a Battle-Tested Open-Source Agentic RAGgeneral2026-05
Email Text Classification Under Severe Class Imbalancemanufacturing2026-05
A Synthetic Training Pipeline for Medical LLMshealthtech2026-05
Analyzing Public Pedagogical Benchmarks for AI Tutors and Building Synthetic Students for Testingedtech2026-05
Advancing semantic search for millions of Rovo userssaas-b2b2026-03
Building AI search: from prototype to productionsearch-engine2026-01
Exploring How to Improve Assessment with AIedtech2025-10
Building the next generation of job search at LinkedInmarketplace2025-05
Retrieval-Augmented Fill-in-the-Middle: An Architecture for Generating Structured Documents with LLMsgeneral2025-05
Embedders 2025: A New Vector in Text Representationretail2025-05
Function Calling 101: One Agent Instead of a Zoo of Bots—general2025-05
Solving Data Discovery at Scale: How Wix Uses RAG and Multi-Agent Systems to Find the Right Data Fastsaas-b2b2025-05
How to improve search without looking at queries or resultssaas-b2b2024-11
How Can OpenAI Help If You Use Open-Source LLMs?edtech2024-11
Advanced Use of GigaChat at Rabota.ru: Synthetic Data and Searchclassifieds2024-05
RAG Evaluation. Evaluate yourself. How to teach LLM to evaluate LLM workretail2024-05
RAG as a Flexible Method for Creating Language Assistantsgeneral2024-05
An LLM for a Tax Agentgeneral2024-05
How We Got RAG Running in Productiongeneral2024-05
Review and Adaptation of RAG Evaluation Tools for the Russian Languageinsurance2024-05
Building a Playlist You Will Likemedia-streaming2024-05
Phone Number Scanner—banking2023-05
LLaMA and a Zillion Ways to Fine-Tune It—general2023-05
Generation of three-dimensional synthetic data for training neural networksgeneral2022-10
Introducing Natural Language Search for Podcast Episodesmedia-streaming2022-03