Speaking
ML Conference, Munich and Berlin
Sessions on document understanding at scale, context engineering and reliable agents; Munich 2026 completed, Berlin 2026 upcoming, three earlier sessions in 2025.
Speaking →Engineer · Founder · Builder · Speaker · Author
I build to understand.
I’m Sinda Khenine, an engineer, founder, speaker, author and builder. My work moves between data, AI, autonomous agents, products, interactive experiments and writing. The subjects change, but the instinct stays the same: understand something deeply enough to build something useful from it.
I spend my days on enterprise data platforms and reliable AI systems. I spend the rest of my time asking what those systems could become: a company run by agents for 24 hours, an arena where agents face consequences, a story you can walk into, a rehabilitation week with less friction, and a book about the structures people carry without noticing. Different subjects, one habit.
02 · The thread through my work
Every project started as a question I could not answer at a desk.
Different projects come from different questions. A platform, an experiment, a prototype and a story do not look alike, but each one exists because reading about the subject was not enough. At some point the only way to understand it was to build it and watch what happened.
- SKatalyst AIHow can fragmented organizational information become structured context and useful software?
- One-Person AI CompanyHow much ongoing work can actually be delegated to persistent agents?
- RehaHow can technology reduce friction around a real human routine?
- Grok ArenaWhat changes when AI agents are given roles, tools and consequences?
- The Magical BookCan generated media become an experience someone feels they entered?
The answers live in the case studies. What this page is about is the habit behind them: curiosity, and building things to understand difficult systems.
03 · Professional foundation
Underneath the experiments is an engineering career.
More than eight years across industrial analytics, machine learning, data platforms and AI-driven decision systems. The experiments borrow their discipline from this work: evidence, contracts, review, and systems a real team can maintain.
- 2020 to presentSenior Big Data & Business Intelligence Engineer
Data platforms, analytical decision systems, machine-learning workflows and AI automation, industrial scale.
- 2017 to 2020Data Scientist, then Big Data Analyst
Forecasting, heterogeneous-source integration, consumer-feedback analysis and predictive modelling.
- 2018 to 2020Engineering degree, Big Data & Business Intelligence
Ecole Polytechnique Internationale, after computer-science engineering studies at ESPRIT.
SKatalyst AI is where this foundation became a product: a founder-led platform, in private beta, that turns fragmented business data into software customers own. It is one important chapter, not the whole story. A PhD on reliable enterprise AI, context engineering, data governance and agent systems is planned for late 2026.
- Data engineering
- Large-scale data platforms and pipelines in an industrial environment.
- Big data
- Apache Spark and distributed processing for analytical workloads.
- Business intelligence
- Governed analytics, dashboards and decision systems people actually use.
- Machine learning
- Forecasting, predictive modelling and statistical analysis since 2017.
- AI systems
- Context engineering, document understanding, validation and evaluation.
- Agent systems
- Bounded tools, evidence, human escalation and graduated autonomy.
- Technical architecture
- Deployment choices that keep systems inspectable and portable.
04 · Beyond software
Systems aren’t only technical.
The curiosity that makes me take a data platform apart also applies to people. Speaking is where I test whether an idea survives explanation. Writing is where I find out whether I understood it at all. And the book asks the same question about human behaviour that I ask about software: what hidden structure is producing this pattern?
Writing
Articles and conference work
Drafts on context engineering, governed output and why reliable agents need boundaries, alongside the material behind the talks.
Writing →Author
The Subconscious Architect
A Living Blueprint with five layers beneath behaviour and identity, written for people who want depth without performance. In publication preparation, coming 2026.
The book →In one line