About Datalume
Education That Treats the Work as the Point
Datalume was built around a straightforward idea: the best way to learn AI development is to spend most of your time actually doing it.
Back to HomeOur Story
Where Datalume Came From
Datalume started in 2019 in Ratchathewi, Bangkok, founded by a small group of practitioners who had spent years working on data and ML projects across Southeast Asia. The common thread in their experience was that people who learned well did so by working through real problems — not by reading about them.
The early programmes were informal: weekend sessions at a shared workspace, working through datasets together, reviewing each other's code. Word spread among engineers and analysts in Bangkok who were looking for something more structured than self-study but less abstract than academic coursework.
By 2021 the programmes had taken their current shape — three distinct tracks covering the main layers of practical AI development, each with mentored feedback at the centre. The name Datalume reflects that original intent: light through data, understanding built by working with it directly.
Our Mission
What We Are Here to Do
Datalume's mission is to make practical AI development skills accessible to people in Thailand and across the region — not through introductory slides or theoretical overviews, but through structured work on real data with experienced mentors providing ongoing feedback.
We focus on three areas that we know well: building and evaluating machine learning models, constructing the data pipelines that feed those models, and supporting learners through a substantial independent project. Each programme is narrow enough to go deep and long enough to develop real competence.
Depth over breadth in every programme
Feedback from people who do this work professionally
Portfolio outcomes that show what you can actually do
The Team
People Behind the Programmes
Nattapong Kositchai
Programme Director
Former ML engineer with eight years building production models in finance and logistics. Oversees curriculum design and mentor quality across all three tracks.
Siriporn Wattanasuk
Data Engineering Lead
Specialist in large-scale data pipelines and ETL architecture. Developed the Data Engineering for AI track from the ground up based on real project experience.
Achara Tangkaeo
Mentorship Coordinator
Manages the Capstone Mentorship programme and learner relations. Background in applied research and technical communication across Thailand and Singapore.
Standards
How We Maintain Quality
Mentor Review on Every Submission
No automated scoring. A mentor reads each piece of work and provides written comments before the learner moves to the next stage.
Structured Curriculum with Clear Milestones
Each programme follows a defined learning path. Learners know exactly what is expected at each stage and what they need to produce to progress.
Data Privacy in Practice
All datasets used in programmes are either publicly available or fully anonymised. Learner data and submitted work are stored securely and not shared with third parties.
Open-Source Tooling Only
Programme work uses freely available Python libraries and platforms. Learners keep every piece of code they write and can continue working with it after the programme ends.
Small Group Sizes
Cohorts are kept small so mentors can give each learner adequate attention. We do not expand groups beyond the point where individual feedback quality would suffer.
Regular Curriculum Review
Programme content is reviewed twice a year to reflect changes in tooling and industry practice. Learner feedback is incorporated before each new cohort begins.
Values and Approach
What Shapes the Way We Work
Datalume operates on a small number of convictions that shape every programme decision. The first is that competence in AI development comes from handling real data and real problems — not from completing courses that simulate them at a distance. Our programmes are structured around this from the start.
The second is that feedback from experienced practitioners is irreplaceable. Automated systems can flag certain errors but they cannot tell you whether your modelling choices were appropriate for the data you had, or whether your pipeline structure would hold up at scale. That requires a person who has done the work.
We are also realistic about timelines. Developing sound ML and data engineering skills takes months of consistent effort, not days. Our programmes are designed to be completed properly rather than quickly, and we set expectations accordingly.
Based in Ratchathewi at the heart of Bangkok, Datalume draws learners from across the city and from other parts of Thailand. Remote participation is fully supported. The local community aspect — shared sessions, peer discussion, working alongside others at similar stages — is one of the things learners consistently find valuable.
Next Step
Talk to Us About Your Background
We're happy to have a short conversation about where you are and which programme would make the most sense as a starting point. There's no pressure to enrol until you're confident it's the right fit.
Get in Touch