Personal project
An NLP pipeline for isiNdebele
Building natural language processing for a low-resource language, from scraping text to semantic analysis.
- Python
- BeautifulSoup
- scikit-learn
Hi, I’m Sisekelo.
I help teams turn AI from a pilot into a habit.
I’m an AI adoption consultant in Belgium. I work with organisations and the people in them to find where AI genuinely helps, test it on real work, and make it part of how things get done.
About
Adoption is mostly a people question with a technical side. I came to AI through people development, business, linguistics and data, so I care as much about confidence and habits as about the tools.
What I do
Five things I bring to a team that wants AI to stick.
Adoption programmes
Plans that start from how teams actually work: surveys, maturity conversations and team-by-team sessions, followed through with pilots, owners and clear measures of success.
Finding the right use cases
Turning broad ambitions and everyday frustrations into small, testable use cases with agreed acceptance criteria. Sometimes that means saying “not yet”.
Practical builds
Proofs of concept and automations with the tools a team already has, such as Microsoft 365 Copilot, Power Automate, SharePoint, Azure, ChatGPT and Claude.
Learning and onboarding
Workshops, exercises and onboarding journeys for hesitant beginners and confident builders alike, with responsible-use guidance built in from the start.
Delivery and continuity
Trackers, reporting, documentation and handovers, so the work keeps running when the project team moves on.
Where is your team with AI?
Pick the one that sounds most familiar.
A few people experiment on their own. Others are wary. Nobody is quite sure what is allowed, or where AI would make a real difference.
- Listen first. Short surveys and conversations to understand habits, concerns and where time is lost.
- Choose a clear starting point. A shortlist of use cases rooted in daily work, small enough to test in weeks.
- Set guardrails people understand. Plain-language guidance on what is fine, what needs a human check, and who to ask.
Licences are in place and some pilots are enthusiastic, but use is uneven and the value is hard to show to the people who fund it.
- Different paths for different people. Practical support for beginners and deeper sessions for builders, instead of one training for everyone.
- Reinforce after the workshop. Follow-up exercises, ambassadors and an easy place to ask questions.
- Agree what success looks like. Simple measures per pilot, tracked so decisions rest on evidence.
Some uses clearly work. Now they need to move from a few enthusiasts into normal processes, without depending on one person.
- Turn wins into patterns. Reusable workflows and templates that other teams can adapt.
- Build it into the organisation. AI in onboarding, support channels and existing HR and IT processes.
- Document and hand over. Clear ownership and documentation so the solution outlives the project.
How I work
The same five steps, whether it is one workflow or a whole organisation.
- 1
Listen
Understand the work as it really happens, not as the process chart says.
- 2
Scope
Agree a bounded first outcome and what “done” means.
- 3
Build
Test with real material, alongside the people who will use it.
- 4
Validate
Check results honestly, with a human in the loop where it matters.
- 5
Hand over
Document, train and make sure it keeps working without me.
My projects
A few things I built while moving into data and AI. Each one has its own write-up.
BeCode company project with Orange
Recommending subscription plans with machine learning
A recommender system for subscription plans that reached 80.9% accuracy on unseen data.
- Python
- scikit-learn
- PyCaret
- XGBoost
BeCode company project with Accenture
Wine market analysis in Tableau
Finding trends and patterns in wine market data to guide marketing and sales decisions.
- Tableau
- SQL
How I got here
Each step taught me the same thing: a new tool only matters once people can use it with confidence.
- 2025 to now
AI Consultant, Buildwise
AI adoption, use-case discovery and practical proofs of concept across the Belgian construction sector, inside the organisation and with companies.
- 2024
Junior Data Analyst Traineeship, BeCode
Seven months of hands-on data work: Python, SQL, machine learning, NLP and dashboards in Power BI and Tableau.
- 2022 to 2023
Learning and Development Officer, Deloitte
Helped bring onboarding and learning data into one platform, and made reporting across teams more consistent.
- 2021
MA African Studies, Linguistic analysis, Ghent University
Where a fascination with how language shapes understanding became a discipline.
- 2018
BSc Business Administration (ICT), KU Leuven
Followed by market research and customer analysis as a Sales Ambassador at JIVR.
- 2014 to 2016
Country Manager, AIESEC in Belgium
Led training and coaching for 50 team leaders and mentored more than 300 student volunteers.
From the blog
Notes on careers, learning, language and identity.
-
Writing
I wrote a book about identity
On writing My Identity, My Dreams, an autobiographical exploration of identity.
-
Career
Crafting a data-focused career from diverse experiences
How work in HR, sales, marketing and research fed into a career built around data.
-
Learning
Becoming a data scientist in 7 months
Seven intensive months at BeCode, and what changing careers taught me about learning.
Let’s talk about AI that people actually use.
The best way to reach me is a message on LinkedIn. I’m always happy to swap notes on AI adoption, learning and change.