Applied AI research & experimentation
Turn a business question into a testable product direction.
Applied AI research & prototyping
Applied AI systems, from research to working prototypes
We help early-stage teams identify useful AI opportunities, design and test LLM and agent-based workflows, and understand where those systems succeed or fail.
01 / About
Useful AI, carefully tested.
AllofMines researches, prototypes, and evaluates practical AI systems. We work where business problems meet technical reality—making ideas concrete, testing their assumptions, and documenting the failure modes that matter.
02 / What We Do
From a promising question to evidence.
LLM & agent-based prototypes
Design and test practical language-model and agent workflows around real constraints.
Data & knowledge workflows
Build curation, annotation, and knowledge systems that make AI inputs more useful and inspectable.
Strategy, evaluation & failure analysis
Study feasibility, define evaluation plans, and understand where systems succeed or break.
03 / Selected Work
EdgeIMCI — Clinical AI, Speech Data & Evaluation
Selected contributions to a research and product effort exploring speech data, clinical information extraction, and evaluation workflows. AllofMines does not own EdgeIMCI, and this work should not be read as clinical validation or deployment.
Developed a consent-based speech-data platform for Nigerian English, Nigerian Pidgin, Yoruba, Igbo, and Hausa, including language and code-mixing metadata.
Collected 380+ accepted recordings and designed transcription, annotation, and quality-controlled evaluation workflows.
Built ASR benchmarks tracking recognition accuracy, latency, and failure cases, plus voice-input review and correction workflows for extracted clinical findings.
Fine-tuned and evaluated a small language model for structured clinical-information extraction, using downstream decision-equivalence tests to identify limitations.
04 / Writing
Technical Notes from Niniola.
Writing on AI, technology, and philosophy.
05 / Contact
Have a problem worth testing?
Tell us what you are exploring. We can help frame the opportunity, prototype a path, and evaluate what happens in practice.