Open for applications · January 2027 intake
Data Science to AI Engineering
A project based track from your first line of Python to production AI, built on fraud, credit and M-Pesa data from Kenyan financial services. Five months full time or eight months part time. Leave with a public portfolio, deployed work, and a clear path into data and AI roles.
5 or 8
months, full or part time
288
live contact hours
24
projects you ship
0
coding experience required
- Admission status
- Open for applications
- Next intake
- Starts 11 January 2027
- Duration
- 5 months full time or 8 months part time
- Delivery
- Hybrid: online on weekdays, in person on Saturdays
- Fees
- KES 140,000, payable in instalments
- Entry requirements
- 18 and above. No coding or data experience needed
- Certificate
- Mctaba Labs certificate on completion
- Facilitated by
- Google Cloud Certified ML Engineers
Career paths
Roles this prepares you for
Most programs stop at analyst. This one runs from analyst through data science, MLOps and AI engineering, so you leave with more than one door open. These are the roles the track prepares you for at entry to mid level, and what each one actually does.
Ready from Week 6
Data Analyst
Answers business questions with SQL, Pandas and charts, cleans and validates data, and writes the one page memo a manager acts on.
Ready from Week 6
BI and Reporting Analyst
Builds the Power BI dashboards, KPI definitions and data models leadership reads every morning, and keeps them honest.
Ready from Week 13
Fraud and Credit Risk Analyst
Scores transactions and loan applicants, tunes decision thresholds against real cost, and documents decisions a regulator can read.
Ready from Week 16
Junior Data Scientist
Frames the problem, engineers features, trains and evaluates models for credit, churn and fraud, and explains them with SHAP.
Ready from Week 21
Junior Machine Learning Engineer
Packages models as APIs, sets up CI, monitoring and retraining, and keeps models alive after launch, including on live transaction streams.
Ready from Week 24
Junior AI Engineer
Builds LLM features with retrieval, evaluation sets and guardrails on top of a company's own documents and data, and watches the token bill.
Also a fit for:
- Product Analyst
- Operations Analyst
- Analytics Engineer
- MLOps Engineer
- Data Science Consultant
Where the work is
Four ways graduates get hired
We prepare you and we open doors. The offer is earned by your portfolio and your interviews, which is why both are built into the program rather than left to chance.
Kenyan financial services
Banks, fintechs, SACCOs, insurers and telcos hire analysts, risk and data teams. The whole track is built on their problems, so you interview with relevant work in hand.
Remote roles with international teams
Data and ML teams abroad hire remotely across time zones. A public GitHub, deployed demos and published articles are how a Nairobi candidate gets noticed.
Freelance and contract work
Dashboards, data clean ups, forecasting and small model builds for SMEs and NGOs, found through your network and freelance platforms.
Startups and studios
Early data hires at startups, and placements on Mctaba Labs client projects for strong performers, where you keep building after demo day.
Who it is for
Built for five kinds of learner
Learners are 18 and above. The first two weeks teach Python from zero and math is taught through code in Week 7, so no background in either is assumed.
Complete beginners
Entering a data career from scratch and need a gentle start with real projects from Week 1.
Recent graduates
You need SQL, deployment and a public portfolio that gets you placed, not more theory.
Finance and ops professionals
You know the business. You need Python, ML and MLOps to match that domain knowledge.
Developers and IT staff
You can code. You want statistics, model evaluation, MLOps and AI engineering.
Self taught learners
You have watched the tutorials. You need structure, feedback and a cohort to finish.
The curriculum
Nine phases, from beginner to advanced
Timed checkpoints at the end of Weeks 6, 13 and 21 confirm you are ready for the next level. The team capstone runs in Weeks 23 and 24.
Phase 01Weeks 1 to 2
Intro to Data Science and Python
Python from zero, Git and GitHub, APIs, scraping, regex and tests. Build a Pesa Tracker and a Live Rates Bot.
Beginner
Phase 02Weeks 3 to 6
Data Handling, EDA and Business Intelligence
OOP, SQL, Pandas, data cleaning, EDA, Power BI and Streamlit. Ends with Checkpoint 1 and a public Spending Insights App.
Beginner to Intermediate
Phase 03Weeks 7 to 8
Math, Statistics and Time Series
Linear algebra and calculus through code, gradient descent from scratch, Bayes for fraud alerts, A/B testing and time features.
Beginner to Intermediate
Phase 04Weeks 9 to 13
Machine Learning
Linear and logistic models, metrics for imbalanced fraud data, XGBoost and LightGBM, clustering, SHAP and fairness. Ends with Checkpoint 2.
Intermediate
Phase 05Weeks 14 to 16
Deep Learning
Neural networks in PyTorch, CNNs and OCR for KYC documents, forecasting, NLP and fine tuning a transformer on Hugging Face.
Intermediate to Advanced
Phase 06Weeks 17 to 18
MLOps and Production AI
MLflow, Docker, FastAPI, CI/CD, Cloud Run, monitoring for drift, cloud pricing and cost efficient fine tuning with LoRA.
Advanced
Phase 07Weeks 19 to 21
Real Time and Online Machine Learning
Kafka streams, concept drift, online learning with River, Vowpal Wabbit, bandits and drift detectors. Ends with Checkpoint 3.
Advanced
Phase 08Week 22
Intro to AI Engineering
LLM APIs, prompting, retrieval augmented generation, agents, evaluation sets and guardrails for a financial assistant.
Advanced
Phase 09Weeks 23 to 24
Capstone and Career
Build an end to end financial ML product with a team, deploy it, then present at a public demo day with guests and employers.
Advanced
What you will build
A visible win every single week
Twenty four mini projects, each finished in class, pushed to GitHub and graded on a shared rubric. Here are six of them.
Week 6
Spending Insights App
A public Streamlit app and Power BI report presenting three insights from a financial dataset. Send the link to a friend and they click through it on their phone.
Week 10
Fraud Cost Slider
Drag a decision threshold and watch shillings saved move against customers wrongly blocked. Metrics chosen by business cost, not accuracy.
Week 13
Why Was I Declined?
Pick a loan applicant and see a SHAP chart of exactly why they were scored that way, with plain language reason codes a regulator would accept.
Week 15
Snap a Receipt
Photograph a receipt with your phone and get the merchant, date and total back as structured data. Document AI with a privacy checklist.
Week 18
Cost vs Quality Scorecard
A green CI badge, a live Cloud Run URL, a monitoring dashboard and a one page comparison of accuracy, latency and cost across three approaches.
Week 22
Chat With a Financial Document
A RAG app over a bank annual report that answers questions with the source page cited, plus an evaluation set that catches hallucinations.
Capstone project
One financial ML product, built with a team
Choose M-Pesa fraud detection, credit scoring, customer segmentation or a RAG support assistant. Scope it, baseline it, build it, deploy it and present it live.
What you build
A data pipeline with quality checks. A baseline and a tuned model with error analysis. A FastAPI service in a container. Monitoring, a cost estimate and a model card. A Streamlit or Power BI front end.
What you hand in
Team design document. Data and model card with fairness checks. Complete GitHub repository with README and architecture diagram. Deployed API and dashboard. Demo script and a live presentation on demo day.
How the program runs
Four sessions a week, full time or part time
Same 24 module curriculum and the same 288 live hours either way. Weekday sessions are online, Saturdays are in person in Nairobi, and every session is recorded.
| Full time | 5 months · 4 hours a session |
|---|---|
| Part time | 8 months · 2 hours a session |
| Weekdays | Live online, recorded |
| Saturdays | In person, Nairobi |
| Every session | Quiz, live coding, paired lab, challenge |
Tools you will use
- Python
- SQL and PostgreSQL
- Git and GitHub
- Pandas and NumPy
- Power BI
- Streamlit
- Scikit-learn
- XGBoost and LightGBM
- SHAP
- PyTorch
- Hugging Face
- MLflow
- Docker
- FastAPI
- GitHub Actions
- Google Cloud Run
- Kafka
- River
- Vowpal Wabbit
- LangChain
- Chroma and FAISS
- Claude, OpenAI and Gemini APIs
Labs run on Google Colab and Kaggle, so a modest laptop is enough. Sessions are recorded and datasets are downloadable once, to keep your data costs down.
Outcomes, measured
How we make sure you actually get there
Attendance is not the outcome. Work is. Every week you produce four things an employer can look at, and every milestone is a timed test of whether you can do the job.
Every week
You ship a project
A mini project finished in Session 4, pushed to GitHub, demoed to the class and graded on a shared rubric: correctness, reproducibility, insight, responsible practice, visible result.
Every week
You write it up
A short technical article on dev.to explaining what you built, and a research report on a set topic, so you can defend your choices in an interview, not just run the code.
Every week
You get reviewed
Two assignments with written feedback, code review in pairs, a warm up quiz and an exit ticket every session, and mentors on hand in every lab.
At the milestones
You prove it
Timed practical checkpoints in Weeks 6, 13 and 21, a team capstone, a public demo day with employers, mock interviews with feedback, and a portfolio review.
How the grade is built
- Mini projects 30%
- Capstone and demo day 25%
- Assignments 20%
- Checkpoints 20%
- Participation, articles, peer review 5%
Pass: 70% overall with 80% attendance
Who teaches you
Google Cloud certified machine learning engineers
The track is facilitated by Google Cloud Certified Machine Learning Engineers who have built and shipped models in production, supported by technical mentors in every lab and peer pods that keep you accountable between sessions.
Mctaba Labs is a software and AI development studio first. The people teaching you build real products for real clients, and the curriculum is written from that work.
1:1
Mentor feedback on every mini project
Pairs
Guided labs and code review in pairs, every session
Pods
Small peer pods for accountability between sessions
Alumni
A community and employer showcase after demo day
Admission and logistics
The January 2027 intake
| Starts | 11 January 2027 |
|---|---|
| Duration | 5 months full time or 8 months part time. 24 modules, 288 live hours. |
| Schedule | 4 sessions a week. Full time 4 hours a session, part time 2 hours a session. |
| Delivery | Hybrid. Online on weekdays, in person in Nairobi on Saturdays. Every session is recorded. |
| Fee | KES 140,000, payable in full or in instalments via M-Pesa or card. |
| Requirements | 18 and above. A laptop and internet. No coding or math background needed. |
| Certificate | Mctaba Labs certificate on passing, plus the portfolio that proves it. |
How to join
Apply online
About six minutes, in five short steps.
Intake quiz
A short readiness check and survey.
Friendly call
A relaxed chat with the team about your goals.
Onboarding
Place confirmed, payment plan and setup guide sent.
Apply for January
We will send the intake quiz and confirm your place by email.
Apply for JanuaryQuestions? Write to hello@mctaba.com or WhatsApp +254 727 795 603.
Common questions
I have never coded. Can I keep up?
Yes. Weeks 1 and 2 teach Python from scratch, and every session has a guided lab in pairs plus mentors on hand.
Do you guarantee a job?
No, and be wary of anyone who does. We guarantee the work: a public portfolio, deployed projects, published articles, mock interviews and an employer showcase. That is what gets people hired.
How is this different from a data analytics course?
Analytics is the first six weeks. The remaining eighteen take you through machine learning, deep learning, MLOps, streaming and AI engineering, which is where the senior roles are.
Full time or part time, which should I pick?
Both cover the same 24 modules and 288 live hours. Full time is 4 hours a session over 5 months for people who can commit their days. Part time is 2 hours a session over 8 months and fits around a job.
Can I pay in instalments?
Yes. The fee is KES 140,000 and can be paid in instalments via M-Pesa or card. We confirm the plan with you at onboarding.
I use a Mac or Linux. What about Power BI?
Power BI Desktop is Windows only, so Mac and Linux learners use the Power BI web service or Looker Studio.
Is there a certificate?
Yes, on passing with 70% overall and 80% attendance, plus the portfolio that proves it.
What if I miss a session?
Every session is recorded. Attendance still counts toward the 80% threshold, so plan to show up.
OPEN FOR APPLICATIONS · STARTS 11 JANUARY 2027 · KES 140,000
Apply for the January intake
Apply online at mctaba.com/apply, complete a short intake quiz, and we confirm your place and send onboarding details before 11 January.