English (Ireland) (en-IE)Gaeilge Acadúil na hOllscoile (ga-IE)
IRISH ACADEMIC RESEARCH PODS • MATHEMATICAL SCIENCES & QUANTITATIVE SYSTEMS · Ireland (IE)

Portfolio Theory for the Agentic Firm at Ulster University (Northern Ireland)

Grounded in your lectures, tutorials, and problem sheets across BSc (Hons) in Data Science & MSc in Data Science (With Applied Industrial Placement) at Ulster University (Northern Ireland), our research fellowship bridges rigorous modern probability and optimisation with production-grade agentic systems. Rather than treating mean-variance efficiency or high-dimensional statistics as textbook abstractions, you will deploy 100% of your university mathematics directly into live multi-agent execution graphs where correlation is the new headcount. Structured as an asynchronous, part-time commitment of 10–25 hours per week, these remote R&D pods slot cleanly alongside your degree timetable, Michaelmas and Hilary term deadlines, and lab schedules. You will collaborate alongside fellow Irish mathematicians, computing scholars, and quantitative researchers to publish reproducible preprints and solve the fundamental stability limits of autonomous LLM collectives.

University: Ulster University Part-Time: 10–25 hrs/wk · $20–$45/hr USD Exam-Safe: 0 hrs/wk Pause Primary Pod: POD-02 Vega-48 Fit: 97.67/100
View Pod Roster (Baserow) → Apply for Research Pod Access Research Preprints Download Seminar Lab Kit
01 // STUDENT PART-TIME PILLAR

Asynchronous Modularity: 10–25 Hours/Week Fitted to Your Timetable

Engineered purposefully around Irish university terms, lecture timetables, and tutorial blocks. Research pod contributions are fully asynchronous, enabling you to coordinate work flexibly around weekly problem sheets, lab sessions, and study commitments without mandatory daytime meetings.

02 // STUDENT PART-TIME PILLAR

$20–$45/hr USD Compensation & Strict Exam-Safe Protocol

Earn $20 to $45/hr USD in remote compensation based on research experience and mathematical seniority. Under our strict Exam-Safe Protocol, all pod obligations pause completely to 0 hrs/week during study weeks, midterms, and winter/spring examination diets, with your seat and research workstream unconditionally guaranteed upon return.

03 // STUDENT PART-TIME PILLAR

Thesis Synergy, Peer Preprints & 39-Hour CPE Library

Directly connect your pod experiments with your Senior Sophister honours dissertation, MSc thesis, or doctoral research projects. Co-author peer-reviewed 2024–2026 preprints and access our comprehensive 39-hour Continuous Professional Education (CPE) research repository at {preprints_url}.

100% UNIVERSITY COURSEWORK → PRODUCTION PORTFOLIO MATHEMATICS

Ulster University (Northern Ireland) — BSc (Hons) in Data Science & MSc in Data Science (With Applied Industrial Placement)

Academic Module & Subject Mathematical Formulation R&D Pod Workstream Applied Systems Implementation
Convex Optimisation & Quadratic Programming min_w w^T Σ w s.t. w^T μ ≥ μ*, 1^T w = 1, w ≥ 0 POD-01: Agentic Mean-Variance Allocation Allocating execution compute weights across heterogeneous LLM and deterministic verifier agents by formulating and solving constrained quadratic optimisation programs under strict latency-cost budgets.
Probability Theory & Correlated Random Processes N_eff = N / (1 + (N - 1)ρ̄) POD-01: Correlation & Ensemble Orthogonalisation Proving why 50 correlated prompt replications at average pairwise correlation ρ̄ = 0.70 yield an effective sample size of only N_eff ≈ 1.42 independent experts, and applying Gram-Schmidt prompt orthogonalisation to restore ensemble diversity.
High-Dimensional Mathematical Statistics & Econometrics Σ̂_LW = (1 - δ*)S + δ*F POD-02: Covariance Shrinkage & Spectral Denoising Regularising ill-conditioned empirical covariance matrices when evaluating LLM agent collectives over short benchmark evaluation runs (N > T) using Ledoit-Wolf shrinkage and Marčenko-Pastur spectral filtering.
Actuarial Risk Theory & Extreme Value Analysis Σ⁻_ij = E[min(R_i-τ,0)min(R_j-τ,0)], CVaR_99%(L_w) ≤ M_audit POD-02: Asymmetric Risk & Tail Failure Bounds Quantifying catastrophic downside hallucination risks via Markowitz 1959 downside semivariance and constraining 99% Conditional Value-at-Risk within deterministic verification budgets via Rockafellar-Uryasev optimisation.
Financial Economics & Stochastic Calculus w_tan ∝ Σ⁻¹(μ - c·1), SR(w) = w^T(μ - c) / √(w^T Σ w) POD-03: Capital Allocation & Tangency Architectures Operationalising James Tobin's separation theorem: treating deterministic compiler checks and formal verifiers as the zero-variance risk-free asset (σ_0 = 0) and synthesising the tangency portfolio of stochastic LLMs to maximise the information Sharpe ratio.
Sequential Hypothesis Testing & Applied Statistics P(Type I Error) ≤ α (Neyman-Pearson FrugalGPT/RouteLLM Cascade) POD-04: Sequential Cascades & Research Preprints Deploying Neyman-Pearson optimal routing cascades across frontier and lightweight LLM tiers under guaranteed Type I error bounds; co-authoring verified empirical findings through our open 39-hour CPE research library at {preprints_url}.
STUDENT-READABLE CAMPUS CHANNELS · MyCareer (TCD / UCD / DCU) & GradIreland Quant Board · TCD Mathematical Society, DUCSS & UCD Actuarial/Quant Society

Campus Engagement & Research Channels at Ulster University (Northern Ireland)

1. MyCareer (TCD / UCD / DCU) & GradIreland Quant Board: To formalise your application for an accredited internship, research fellowship, or departmental co-op placement, log into MyCareer (TCD / UCD / DCU) & GradIreland Quant Board with your official Ulster University (Northern Ireland) student single sign-on. Filter active listings under 'Quantitative Research & Mathematical Systems', verify your module prerequisites with your academic tutor or placement coordinator, and submit your CV alongside your academic transcript and GitHub handle. (RSS/XML Job Feed · iCal Open Office Hours · Google Classroom)

2. Python Seminar & Thesis Kit (/uni/ie-ulster-university/lab.py): A self-contained, reproducible 5-minute Jupyter notebook and Python CLI script designed for university departmental seminars and problem-solving workshops. It simulates multi-agent LLM response correlation matrices, computes effective sample size N_eff across prompt regimes, and applies Ledoit-Wolf shrinkage Σ̂_LW alongside Marčenko-Pastur spectral clipping to construct numerically stable minimum-variance agent ensembles.

3. TCD Mathematical Society, DUCSS & UCD Actuarial/Quant Society:
Calling all maths, stats, quantitative, and computing students at Ulster University (Northern Ireland)! Markowitz Systems is recruiting for remote, part-time R&D pods ($20–$45/hr USD, 10–25 hrs/wk) exploring 'Portfolio Theory for the Agentic Firm'. If you've ever wondered why chaining 50 correlated LLM prompts collapses your effective sample size to barely one independent expert, this fellowship applies rigorous portfolio mathematics—from quadratic programming to Ledoit-Wolf shrinkage—directly to live multi-agent execution graphs. The appointment includes a strict Exam-Safe Protocol (0 hrs/wk pause during study weeks and exam diets with guaranteed seat retention) and direct synergy with final-year honours dissertations. Review the research briefs and submit your application at https://markowitz.systems/uni/ie-ulster-university.
DIRECT STUDENT R&D APPLICATION → CLOUDFLARE D1 & BASEROW

Apply for Research Pod — Ulster University (Northern Ireland)

View Pod Roster (Baserow) ↗