Technology leadership
From technology strategy to execution, we help teams make clear architectural choices, build delivery discipline, and connect investment to measurable outcomes.
About aixtract
Technology that makes
intelligence accountable.
aixtract helps organizations turn data, documents, and AI outputs into reliable systems of action.
We combine senior technology leadership, applied AI, and traceable workflows to make complex decisions easier to understand, validate, and own.
The practice
Built at the intersection of AI, data, and delivery.
aixtract is founded by Dr. Daoyuan Li, a CTO, Data & AI leader, author, and machine-learning researcher based in Luxembourg.
Daoyuan has led technology strategy, product delivery, infrastructure, data platforms, and security operations for global-market SaaS businesses. He has built and scaled R&D organizations from a single engineer to more than 50 people, and delivered AI and digitization initiatives from first architecture through production use.
His work spans large-scale data systems, LLM and SLM applications, NLP, entity resolution, information extraction, machine translation, image understanding, search, and MLOps. The common thread is practical: advanced technology should improve the quality, speed, and accountability of real decisions.
He is also the author of Cheap Thinking, an examination of what AI makes abundant, what it makes scarce, and who captures the difference.
By the numbers
15+
Years building technology, data, and AI systems50+
R&D team members led and scaled200M+
Companies represented in a proprietary data platform100M+
Records processed in entity-resolution work50
Languages supported by a translation service40–60%
Reduction in project delivery time and costWhat we bring
The discipline to make AI work in the real world.
AI-native product development
We design and ship products that use machine learning and language models where they create durable value, not merely where they create novelty.
Data foundations
Reliable AI depends on reliable data. We build data platforms, governance, annotation, moderation, enrichment, search, and entity-resolution capabilities that hold up at scale.
Traceable workflows
AI outputs become useful when people can inspect the evidence, understand the decision, assign ownership, and monitor what happened next.
Infrastructure and operations
We work across cloud, bare metal, hybrid environments, MLOps, security, observability, and automation so promising systems can become dependable services.
Selected work
Built for scale. Designed for use.
Enterprise data platforms
Architected a proprietary database covering more than 200 million companies, with data KPIs doubling year over year for five consecutive years.
Entity resolution at scale
Optimized an entity-resolution and data-deduplication service across more than 100 million records, achieving a 100× speed-up and 10× broader recall coverage.
AI-powered enrichment
Led LLM integration and fine-tuning for conversational AI and automated data enrichment, including extraction from tens of millions of business websites.
Translation infrastructure
Developed an LLM-based machine-translation service supporting translation from 50 languages into English.
Classification services
Engineered an industry-classification service from business descriptions, handling millions of requests per day at peak.
Search and image intelligence
Built deduplication, ranking, annotation, search, and recommendation services across millions of hotel products, reviews, and images.
How we think
Operating principles.
Make the invisible inspectable.
Important decisions should expose their evidence, assumptions, and limitations.
Design for the whole system.
Models are only one component. Data, workflows, infrastructure, controls, people, and incentives determine the outcome.
Prefer useful complexity.
The best architecture is the one that creates durable value and can be operated responsibly.
Move from experiment to ownership.
A prototype becomes valuable when someone can run it, measure it, improve it, and be accountable for its consequences.
Work with aixtract
If the decision matters, the system behind it should be inspectable.
aixtract works with leaders who want to move beyond AI pilots and build systems people can trust.
Talk to aixtract