Pillar /03 · Training

Training at AIRINA.

Short-form. Mathematically rigorous. Built around what the unit actually publishes. The training pillar is how AIRINA's research reaches working data scientists, analysts, and graduate students — and how the unit recruits its next research collaborators.

Format 2 days to 2 weeks Languages EN / FR Certificate Graded, defended
What we run

Six programs across three tiers.

The catalog has two anchor programs in Tier 1 (TDA for data scientists, and interpretable ML for regulated industries), two longer programs in Tier 2 (the 2-week summer school on topological foundations of deep learning, and a 1-week quantum-topological methods bootcamp), and two specialized programs in Tier 3 (an asymmetric-topology research school, and a robustness-and-OOD training in TDA). Tier 2 and 3 programs open as the unit's publications in each area mature — we do not teach a topic before we have published on it.

Full catalog with curricula, prerequisites, and registration: training.html →  ·  formations.html (FR) →
How the training pillar works

Four principles.

/01

Short-form, by design

The longest program in the catalog is two weeks. Most are three to five days. We are not building a degree-granting school. We are giving working professionals and graduate students concentrated access to research they cannot get from a Coursera course.

/02

Mathematically deep

The foundation we expect goes beyond Python and sklearn. Calculus, linear algebra, probability, and — for the longer programs — at least one of measure theory or algebraic topology. The unit's edge is mathematical depth. We do not water it down to widen the market.

/03

Graded, defended, real

Every program ends with a project — a TDA pipeline on a real dataset, an interpretability and fairness audit of a real model, a research mini-project — defended in front of AIRINA researchers. No certificate of attendance. The certificate carries weight because the assessment has teeth.

/04

Bilingual, operationally

Technical content is delivered in English (the research language); spoken sessions and written deliverables work in both EN and FR. Participants present in both. Francophone West African analysts and Anglophone graduate students sit in the same cohort.

Who teaches

Practitioners only.

AIRINA researchers and invited faculty from partner institutions (AIMS, ACAS, university math and statistics departments). The director leads the topology-heavy material himself. We do not bring in non-practitioners. If a topic does not have a working practitioner in the unit, we either co-host with someone who does, or we do not run it.

The slow-catalog discipline is intentional. Adding a fifth or sixth program is cheap; teaching it well is not.

Training × research

One engine, two timescales.

Each program has a paid industry-consulting twin. A bank that sends two analysts to TDA for Data Scientists can then hire AIRINA on a six-week engagement applying TDA to its fraud-detection problem. The engagement generates the case study used in the next cohort. The published paper that comes out of the engagement gives the next cohort its credibility.

Top participants from the short trainings are recruited into AIRINA research collaborations on the deeper themes. The certificate program is the front door. The research arm is the back room. Same engine, different timescales, and the work compounds.

Get involved

Three ways in.

As a participant

See the catalog and register: training.html · formations.html (FR)

As a hiring partner

Banks, MFIs, fintechs sending analysts in exchange for first-look hiring rights: contact@airina.africa

As a co-host

Universities and research institutes that want to co-accredit Tier-2 or Tier-3 programs: contact@airina.africa

As a funder

Foundations and development banks underwriting cohort access: contact@airina.africa