I turn business questions into data tools people actually use.
Abraham Anaba · Data & AI Project Manager, software engineer by training.
Projects carried from scoping to go-live.
- Scoping
- Data
- Build
- Go-live
Available now for a data & AI work-study role
I have worked with
- Castor Education
- Université de Douala
- Camerounada
- Cleafun
- STA
- GIT SA
01Selected projects
Three projects, from scoping to go-live.
For each: the problem, my exact role and the main result.
All projectsReducing dropout by steering learning paths with data
- Role
- Data Project Manager (work-study)
Making diplomas tamper-proof with blockchain
- Role
- Blockchain project lead (final-year internship)
- Result
- 6 months from scoping to AWS deployment
Digitising a five-branch network and its customer acquisition
- Role
- IT Consultant (freelance)
- Result
- 5 branches equipped and supervised
02Demonstration
Data is only useful when it changes a decision.
At Castor Education, the question was simple: at what point should someone call a learner? Move the threshold to see the trade-off. The data is synthetic, the method is real.
Learners at risk of dropping out, by alert threshold
At this threshold, the team must contact 46 learners to catch 5 dropouts out of 10.
4 of these alerts concern learners who would have stayed. 29 dropouts slip under the radar.
- Dropouts caught
- Unneeded alerts
- Dropouts missed
- Stayed, not flagged
- Dropouts caught
- Alerts that were right
Click or drag on the curve to pick a threshold.
View the data
| Threshold | Flagged | Dropouts caught | Alerts that were right |
|---|---|---|---|
| ≥ 30 | 225 | 100 % | 32 % |
| ≥ 35 | 185 | 99 % | 38 % |
| ≥ 40 | 149 | 99 % | 47 % |
| ≥ 45 | 116 | 97 % | 59 % |
| ≥ 50 | 87 | 83 % | 68 % |
| ≥ 55 | 62 | 73 % | 84 % |
| ≥ 60 | 46 | 59 % | 91 % |
| ≥ 65 | 31 | 41 % | 94 % |
| ≥ 70 | 20 | 28 % | 100 % |
| ≥ 75 | 15 | 21 % | 100 % |
| ≥ 80 | 11 | 15 % | 100 % |
| ≥ 85 | 5 | 7 % | 100 % |
| Risk score | Dropped out | Stayed |
|---|---|---|
| 0–4 | 0 | 1 |
| 5–9 | 0 | 4 |
| 10–14 | 0 | 9 |
| 15–19 | 0 | 19 |
| 20–24 | 0 | 22 |
| 25–29 | 0 | 40 |
| 30–34 | 1 | 39 |
| 35–39 | 0 | 36 |
| 40–44 | 1 | 32 |
| 45–49 | 10 | 19 |
| 50–54 | 7 | 18 |
| 55–59 | 10 | 6 |
| 60–64 | 13 | 2 |
| 65–69 | 9 | 2 |
| 70–74 | 5 | 0 |
| 75–79 | 4 | 0 |
| 80–84 | 6 | 0 |
| 85–89 | 2 | 0 |
| 90–94 | 2 | 0 |
| 95–99 | 1 | 0 |
03Method
My method fits in four steps.
- 01
Understand the decision
Who decides what, how often, and on what basis today.
- 02
Make the data reliable
Known sources, measured quality, one version of the numbers.
- 03
Build with the users
A first tool within weeks, refined with the people who use it.
- 04
Measure adoption
The project ends when the tool is used, not when it is delivered.
04Writing
Field notes
A school’s question, one Tuesday morning. It wasn’t really about AI. It was about what will still be valuable in education tomorrow.
15 March 2026Scope before you code: what CRISP-DM taught meMost data projects don’t fail on the model. They fail earlier: on the question asked. Why I spend so much time on scoping.
05Contact
Hiring a work-study student?
I am looking for my next data and AI work-study role. Here are the essentials; an email or a few lines here are enough to set up a conversation.
Available now for a data & AI work-study role
- Target role
- Data & AI Project Manager, Data Analyst
- Programme
- MSc Data & AI Project Management, ESIC Paris (RNCP 37137, level 7)
- Schedule
- 3 weeks at the company, 1 week at school
- Start
- Immediately (September 2026)
- Programme ends
- September 2027
- Location
- Paris and Île-de-France
Or leave me a note here
I reply personally within 48 hours.