AI Research Resident - Indicina

Job Title: AI Research Resident
Location: Hybrid (Lagos, Nigeria)
Employment Type: 12-month placement, full-time conversion path

 

About Indicina

We build the decisioning infrastructure that powers Africa's next generation of lenders. Our Decide suite - AI-powered credit decisioning APIs - is used by Nigeria's largest banks and leading non-bank lenders. We have accumulated one of the largest labeled corpora of African bank statement data and credit outcomes in the market. It is a rare research asset. We are now investing in foundational modeling work to turn that data into reusable representations that will power the next generation of Decide.

The Role

We are hiring an AI Research Intern to join our AI Science team and work on transaction-sequence foundation models - learned representations of bank statements and transactional data that can be reused across credit decisioning, fraud detection, and behavioral scoring.

This is a research role with a real product line behind it. You will work directly with our AI Science team on an end-to-end research project: literature review, architecture design, training, evaluation, and - where the work proves out - productionization. This is a 12-month placement; strong performers will be offered full-time conversion.

Key Responsibilities

Survey the state of the art in tabular and sequence foundation models.

Design self-supervised pre-training objectives suited to the structure of bank statement data.

Train transformer-based foundation models on Indicina's proprietary transaction corpus.

Benchmark learned representations against current production models on credit risk, fraud, and behavioral scoring tasks. Carry out research on architecture, sequence length, and tokenization choices.

Document findings in a technical writeup; contribute to a publication or open-source release where appropriate.

Qualifications & Skills

We hire on demonstrated ability, not credentials. We consider PhD, Masters, final-year undergrad, and self-taught candidates equally if the signal is strong.

Strong Python and deep learning fundamentals (PyTorch or JAX). Comfortable training models from scratch - not just calling APIs.

Demonstrated ability to read ML papers and implement them.

Working understanding of transformer architectures, self-supervised learning, and either time-series or tabular modeling.

Strong data manipulation skills (pandas, numpy) and intuition for messy real-world data.

Curiosity, rigor, and willingness to defend and revise your modeling choices in technical discussion.

Bonus: published paper or arXiv preprint; open-source ML contributions; Kaggle competition record; experience with financial, behavioral, or sequence data; familiarity with Hugging Face Transformers, Weights & Biases, or similar tooling.

Why Join Us?

A 12-month placement on a genuine research problem with a proprietary dataset few teams in Africa have access to.

Direct mentorship from Indicina's AI Science team and end-to-end ownership of your project.

Full-time conversion path for strong performers.

Competitive monthly stipend, hybrid work, and the equipment you need.

Apply

Send your CV and a link to something you've built - a GitHub repo, a paper, a notebook, a blog post explaining a paper, a Kaggle solution - that shows us how you think. ML applied to tabular or sequential data is a plus, but any rigorous technical work that demonstrates depth will do.

Send your profile to hr@indicina.co