Medical physics · optimization · operational intelligence

TensorChrono

A production medical physicist workforce platform for constraint-aware geographic scheduling, assignment optimization, predictive analytics, equipment logistics, fairness, travel control, and auditable manager decision support.

Technical leadership

Albert Darkhosh is a PhD physicist and the Principal AI/ML & Software Architect for TensorChrono.

Production optimizer

Stable Geographic Optimizer v2 combines deterministic geographic ranking, hard-constraint eligibility, distance-aware fairness, bounded fixpoint reruns, coverage repair, and final swap/rotation convergence.

Operational scope

Schedule ingestion, home-to-job travel, availability and No Travel enforcement, role skills, workload caps, adjustable gap relaxation, equipment kits, assignment rationale, archives, audits, analytics, and confirmation lifecycle.

Optimization algorithms and advanced technology

TensorChrono exposes a governed portfolio of production, exact, metaheuristic, predictive, reinforcement, and AI-assisted methods. Each method shares final validation and repair, while the application clearly identifies method-specific limitations and fallbacks.

Stable Geographic v2Target-Aware GeographicBalanced Heuristic MILP / CBCOR-Tools CP-SATGenetic Algorithm ALNSIterated Local SearchTabu Search Ant Colony OptimizationCross-Entropy MethodSimulated Annealing Portfolio SearchML Predictive ScoringOffline RL / UCB DRL ResearchLLM-Assisted Planning

Engineering platform

The system uses React and Vite for its workflow interface, FastAPI for its application API, PostgreSQL for operational and analytical state, RabbitMQ for optimization tasks, and Google Cloud Run for immutable service and worker deployment. Predictive components use versioned model artifacts and leakage-aware data splits; optimization runs are auditable through algorithm identity, parameters, cache fingerprints, rationale, and lifecycle records.

Designed for reviewable decisions

TensorChrono treats coverage, availability, role skills, home travel, workload balance, scheduling gaps, manager locks, imported assignments, monthly caps, and equipment readiness as operational policy—not as an afterthought. Managers can compare algorithms, examine travel and fairness diagnostics, apply controlled advice, and confirm only the schedule that should become the official baseline.