Optimizing resources under real-world constraints
A reusable decision system for representing what drives an outcome, simulating possible allocations, choosing the best feasible plan, and learning from what happens next.
open system → $ whoami // REPRESENT // SIMULATE // DECIDE → LEARN $ explore ./systems --group=domain./operational-decision-systems ./governed-interactive-analytics ./private-local-intelligence ./adaptive-analytics-platform ./oxic-nitrogen-fixation ./rubisco-active-learning
James Young
I build decision systems for complex domains. Across decision systems, interactive AI, and scientific machine learning, the same loop turns many interacting variables into a feasible next action. Language models can make that system conversational while reliable calculations, evidence, privacy, and approval boundaries remain explicit.
Make the outcome, choices, context, and limits explicit.
Compare realistic actions before committing resources.
Choose what can be carried out, then use the result to improve the next decision.
Systems in practice
Conversational analytics with answers you can verify
A governed analytics workbench where an LLM interprets the question, tested analytical engines produce the answer, and every result includes its evidence.
open system →Private AI for sensitive, long documents
An offline desktop assistant that keeps documents and model inference on the user’s device while breaking long material into a structured, reviewable analysis.
open system →Analytics that adapt to each organization’s data
A framework that keeps the validated analysis stable while using an LLM to understand each new data environment and create the adapters needed to run it.
open system →Choosing which genes to test for oxygen-tolerant nitrogen fixation
A decision system using scientific machine learning to combine biological evidence and rank gene candidates for scarce experimental time and resources.
open system →Choosing which protein variants to test next
A decision system using scientific machine learning to balance promising protein candidates with experiments that will reduce uncertainty.
open system →