I’m Yannick Bounouar. I build private knowledge assistants, evaluate existing chatbots and connect agents to business systems. We start with the problem your team needs to solve and agree what a working result looks like.
My experience includes AI engineering at the United Nations, model validation at Lyft, Canadian government data platforms, and Gulf telecom analytics through Oliver Wyman.
Remote consulting for teams across the United States, Canada and the Gulf. Delivery in English and French.
Global technology companies, government, international organizations, financial institutions and major Middle Eastern enterprises. Production delivery inside high-trust, high-scale environments.
Selected engagements. Full write-ups on Selected work.
Start with your problem
Choose a service with a clear starting point, a defined scope and a practical handover.
Is your chatbot ready for its next release?
Evaluate an existing assistant with approved questions, grounding checks and repeatable regression tests.
You get: An approved acceptance set, repeatable evaluation runner, failure report and agreed remediation handover.
Explore serviceDo your people need answers from internal documents?
Build a source-linked assistant over an approved document collection, with access tests and a refresh process.
You get: A deployed knowledge assistant with ingestion, retrieval, evaluation examples and an operating runbook.
Explore serviceDoes your agent need to work with a business system?
Connect an existing agent to one business system with typed tools, permissions and explicit action controls.
You get: Typed tool/API code, authentication configuration, action logs, failure tests and an operating guide.
Explore serviceBuild multi-step retrieval, drafting, review and revision workflows with clear human decision points.
Build the ingestion, transformation and cloud foundations your analytics and AI systems depend on.
Need help defining the project? Explore assessments and technical advisory.
Your consulting partner
I combine architecture decisions with hands-on engineering. You work with me to define the scope, review the evidence and deliver something your team can maintain.
M.Sc. Information Systems, WU Wien
B.Sc. Mathematics, Université de Montréal
Knowledge platforms at the United Nations. Cloud data engineering for the Government of Canada. Model validation at Lyft. Telecom analytics through Oliver Wyman in the Gulf.
We name the source collection, release decision or system boundary. Your owners approve the expected behavior and acceptance criteria.
Code, configuration, checks and operating instructions are part of the agreed deliverable. Tool and model choices follow your environment.
Selected work
See the systems I built, the problems they addressed and the work I owned.
Turn fragmented institutional knowledge into governed, reviewable reports and reusable AI access.
Lyft
Replace repetitive visual review with production computer vision that stays controlled through real-world change.
Read the case study →Oliver Wyman, for a Gulf telecom operator
Through Oliver Wyman in Dubai, I built network analytics and models to help a Gulf telecom operator prioritize capacity and 5G investment.
Explore the telecom case study →How we work
Name the problem, the sources or system boundary, and the people who approve the result. Define the deliverables and acceptance criteria together.
Work through representative inputs in your environment. Review working increments or test results, with assumptions and limitations made explicit.
Check the agreed criteria with your team. Transfer the code, results and operating instructions, and record any work that remains.
Bring a document collection, an existing assistant or a system you need to connect. We’ll discuss the problem, the constraints and a useful first scope.