Oksana Sudoma
Independent Researcher / AI Engineer
Technical Project Manager specializing in AI-driven research automation
Independent Researcher / AI Engineer
Technical Project Manager specializing in AI-driven research automation
I'm seeking opportunities in technical project management for AI/ML research, research infrastructure, or AI automation where I can leverage my unique combination of theoretical rigor and practical execution.
With a background in mathematical physics and data science, and now certified as a PMP®, I specialize in transforming complex theoretical challenges into systematic, reproducible, and impactful projects. My expertise lies in orchestrating AI-driven workflows, building robust research infrastructure, and leading cross-functional teams to deliver cutting-edge solutions with precision and agility.
Preprints and publications in mathematical physics, number theory, and complexity theory
O. Sudoma (2025) | DOI: 10.5281/zenodo.17585575
Explicit curvature correction formula with rigorous O(κ²ₐ) error bounds and computational validation achieving 0.59% accuracy on S²_R.
O. Sudoma (2025) | DOI: 10.5281/zenodo.17717905
Subfactor index rigidity forces operator families to be piecewise constant. Discrete jumps emerge from Jones index spectrum below 4, with computational validation (R²=0.98).
O. Sudoma (2025) | DOI: 10.5281/zenodo.17727423
Dimension-dependent depth bounds combining Knill-Laflamme conditions with circuit analysis. Explicit thresholds: D∼d (1D), D∼√d (2D), D∼∛d (3D).
O. Sudoma (2025) | DOI: 10.5281/zenodo.17727423
Geometric measure κ predicts when mixture models are needed (K≥2) vs point estimates (K=1). Existence theorem proven rigorously, validated with ρ=0.671, p=0.001.
O. Sudoma (2025) | DOI: 10.5281/zenodo.17585624
Functorial framework unifying QEC approaches via bounded surjections. Machine-precision validation with kernel geometry explaining code convergence.
O. Sudoma (2025) | DOI: 10.5281/zenodo.17432403
Structured deviations in prime factor count distribution matching Selberg–Delange prediction. Decay |S(x)|/x ~ 1.708(log x)^{-3/2} verified to 6 decimals at x=10⁸.
O. Sudoma (2025) | DOI: 10.5281/zenodo.17616884
The 432→54 cascade explains trace-2 impossibility. First computational enumeration revealing 54 doubly stochastic 3×3 matrices over F₃ with binary trace stratification—traces only in {0,1}, never 2.
O. Sudoma (2025) | DOI: 10.5281/zenodo.17562623
Impossibility theorem: no scalar can be simultaneously isomorphism-invariant and monotone across algorithmic, information-theoretic, dynamical, and geometric complexity. Faithful measurement requires 4D vectors.
AI-driven infrastructure for managing complex research programs at scale
Interactive tool for exploring how prime factors distribute across modular residue classes. Real-time computation with Fourier analysis.
Visualizes omega function patterns, decay laws, and theoretical predictions.
JavaScript · Plotly.js · Number Theory · D3.js
Agent orchestration with state machine workflows, hook management, and Telegram monitoring.
Monitors progress, workflows, experiments launched in your IDE (like VS Code).
Knowledge/Document RAG for your project management and research.
Python · Telegram Bot API · State Machines · Structured Logging · Async/Await
Interactive dashboard demonstrating why you can't average system health into a single number. Proves the scalar impossibility theorem with live visualizations.
Explore DORA metrics, SPACE framework applications. See how scalar averaging hides critical trade-offs.
Vanilla JS · Plotly.js · ES6 Modules · Observer Pattern
Interested in collaboration, research opportunities, or technical program management roles? Let's talk.
