ayodeji — ~/research

cat research.md

# i am interested in ai particularly in mechanistic interpretability and ai alignment.

$ ls active/

ai safety in low-resource languagesactive
non-trivial fellowship

investigating how ai safety properties like alignment, robustness, interpretability degrade in low-resource and tonal language contexts.

mechanistic interpretabilityyorubalow-resource nlpai safety
failure modes of scalable human oversight in large language modelspublished

bridging a decade of research, this review analyzes 70+ papers to identify the nine failure modes currently threatening AI safety.

ai alignmenthuman oversight
read
simple uncertainty signals for hallucination detection in RAG systemspublished

addressing the challenge of LLM hallucinations in RAG systems, this paper explores simple, interpretable uncertainty signals for real-time detection.

ai safetyraghallucinationuncertainty
read
project mosaic: defensive protein-aware screening for benchtop synthesizerspublished
apart research
with abubakar abdulfatah

an open-source protein-aware screening tool that catches adversarial DNA sequences evading standard hamming-distance checks by translating to amino acid homology, defending against context-scrubbed multi-agent LLM biosynthesis attacks.

ai alignmenthuman oversightai safetybiosafety

"your project landed in the top half of all submissions. strong work."

buying safety: a model ai procurement standard for african public sectorspublished
apart research

by embedding strict safety standards into public tenders, african governments can instantly transform ai safety from a passive policy goal into an enforceable market requirement.

ai policyai safety
read

"your project landed in the top half of all submissions. strong work."

comparative analysis of lstm neural networks and arps decline models for production forecasting in niger delta oil wellsaccepted
spe nigeria annual international conference and exhibition (naice 2026)
with michael omijie

research compares LSTM neural networks against Arps hyperbolic decline models for oil production forecasting using the volve field dataset.

aimachine learningneural networksarp's decline modelsproduction forecasting

"your paper has been accepted (naice 2026)"

$ cat interests.txt

—mechanistic interpretability
—machine learning theory
—decision theory
—ai safety
—theoretical computer science

papers and preprints will appear here as they are completed.

in the meantime: send a mail if you want to talk about any of this.