Understanding how people learn, work, and adapt under uncertainty.
A research and engineering laboratory investigating online Bayesian capability inference, structured competency topologies, and adaptive cognitive instruments.
Tracking latent practitioner capability across sparse, non-stationary evidence streams without arbitrary certainty.
Selected Systems & Research
Interactive instruments and computational models pushing the boundary of capability inference.
Capability Inference
Continuous Bayesian belief states that track epistemic uncertainty under sparse evidence, regularizing non-stationary skill decay.
Skills Intelligence
Extracting latent vocational competencies from unstructured experience and mapping tokens onto international ESCO and OFO hierarchical manifolds.
Cognitive Assessment Engine
Computerized adaptive item routing via maximum Fisher information gain, converging to narrow capability bounds in 42% fewer observations.
TVET Labour Market Radar
Empirical quantification of curriculum obsolescence and industrial demand friction across South African vocational trade cohorts.
Synthetic Drift Benchmark
A reproducible simulation harness evaluating Bayesian belief resilience under sudden career transitions versus heuristic moving averages.
Research that moves the frontier.
Probabilistic Capability Inference from Sparse Observation Streams
Moving from static score reductionism to continuous Bayesian belief states that track epistemic uncertainty under sparse evidence. Our filter regularizes non-stationary performance degradation across technical domains.
Skills Intelligence in TVET Ecosystems
Extracting authenticated vocational competencies from unstructured resumes, institutional curricula, and industrial certifications projected onto ESCO and OFO manifolds.
Adaptive Computerized Cognitive Engine
Dynamic next-item routing via maximum Fisher information gain. Tighter capability confidence intervals achieved in 42% fewer observations than static linear batteries.
Areas of Inquiry
Our research unites probabilistic machine learning, psychometric measurement, and empirical labor-market telemetry. Hover each discipline to inspect active directions.
Artificial Intelligence & Probabilistic Kernels
Online Bayesian filtering, conjugate Beta-Binomial state estimation, and entropy-regularized drift compensation under sparse observations.
Data Topologies & Occupational Graphs
Bridging unstructured experience records and international competency frameworks (ESCO/OFO) through semantic token extraction and hierarchical projections.
Cognitive Assessment & Adaptive Instruments
Computerized adaptive testing via maximum Fisher information gain, dynamically self-optimizing evaluation items to tighten capability bounds.
Empirical Labour Market Signals
Validating computational hypotheses against the non-stationary friction of real-world technical colleges (TVETs) and industrial trades in South Africa.
Latest Dispatches
Why conventional skill taxonomies break under sparse evidence
When competency systems rely on keyword extraction without posterior uncertainty bounds, false certainty propagates across hiring manifolds.
Probabilistic Modelling of Latent Human Capability from Sparse Observation Streams
Beta-Binomial conjugate dynamics with temporal drift compensation for tracking practitioner competence in high-friction labor markets.
Adaptive Measurement via Maximum Information-Gain Item Routing
Computerized adaptive testing engine that minimizes assessment burden by 42% while tightening posterior variance.
We build to understand.
We are a research-driven team exploring how technology, data, and intelligent systems can solve meaningful problems under real-world uncertainty.
Rather than testing hypotheses in synthetic vacuum scenarios, we validate our models against the noisy, non-stationary friction of real-world vocational education, technical trade cohorts, and emerging workforce signals across South Africa and the Global South.
