TECHNOLOGY
Intelligence Beyond Inference
Make the world's complexity understandable
Noesis Technology builds AI systems that perceive the hidden structure of markets. The age of information solved access. The real bottleneck now is understanding — and that is the problem we exist to solve.
Noēsis (νόησις) sits at the apex of Plato's hierarchy of cognition — the highest form of knowing. Not perception, not inference, but the direct apprehension of essence.
It names exactly what our technology does: moving past the surface of events to see the causal structure beneath them — to perceive, not merely summarize.
Make the world's complexity understandable.
Markets generate an overwhelming stream of events every day. The signal is rarely missing — what's missing is the connection between scattered events and the assets they ultimately move. We build the engine that draws those connections, with the reasoning path made visible.
Surface the hidden connections others miss.
Deliver the path, not just the conclusion.
Move from information to conviction.
A system that sharpens the more it is used.
In five minutes — with the causal path. Not a news summary, but a reasoning engine: Event → SubImpact → Theme → Stock. Every morning, fund managers scan a few dozen names and miss the other 2,400. Noesis Provenance closes that gap.
An Event → SubImpact → Theme → Stock causal chain, stored and queryable. A reusable library of 50–80 SubImpact blocks means new events only need new connections — the reasoning compounds rather than restarts.
Nine analyst agents reason independently; four supervisors reach consensus — an AI investment committee. An Independence Checker controls for correlated bias, so conclusions are cross-examined, not single-shot.
The current market regime is quantified across six dimensions, always on, and injected automatically into every agent's reasoning. The same event yields different beneficiaries depending on the regime — and the system knows it.
Not a prototype. A live causal graph over the entire Korean market, fed by real-time pipelines every trading day.
Ingestion → causal topology → reasoning → serving → presentation. Events come in; a knowledge graph forms; agents reason; the API serves; the frontend renders.
Retrieval tools find causation that's already written down. They stop at the first-order, already-reported link. Noesis reasons on the causal graph itself — reaching the second- and third-order beneficiaries the market hasn't priced yet.
Founder & CEO. Nine years applying causal-inference methods across four domains — digital pathology, AI drug discovery, industrial robotics, and manufacturing AI — now transplanted into financial markets.
AI Guru — AI Team Lead / Principal Consultant. Manufacturing AI solutions end-to-end, from customer pain points to production deployment. Results include 99% OK-rate and a 67% reduction in downtime.
MakinaRocks — ML Engineer / Chapter Lead. Predictive maintenance for 300+ industrial robots: anomaly detection, XAI fault analysis, Kubernetes & Airflow infrastructure.
DearGen · DeepBio — ML Engineer. AI drug discovery (GNN, BERT, protein structure) and digital-pathology models; patents and publications.
University College London (UCL) — MSc Machine Learning. Distinction, GPA 4.0/4.0.
Imperial College London — BSc Theoretical Physics. First Class Honours, GPA 4.0/4.0.
A founder who pairs research depth with hands-on industrial deployment — turning complex causal structure into decision systems that run in production.
The data and the engine are ready. We're looking for institutional partners to establish the first production reference — and for exceptional people who want to build the causal-inference layer for markets.