Forward-Deployed AI Engineer
I build production LLM systems and the rollout that gets them used — classification pipelines, rubric-based scoring, daily intelligence systems, and the discovery and workshop programme that makes them adopted.
Currently at Volpi Capital, owning the firm's AI/data stack end-to-end. Completing an MSc Computer Science at the University of Birmingham with original research on geometric inference for neural dynamics under Dr Felipe Orihuela-Espina. Six years of prior production systems engineering across KPMG, Siemens, and Aurelle.
Experience
AI Engineer
2026–PresentVolpi Capital
One-person AI/data engineering function for a European lower mid-market PE firm. Shipped four production LLM systems in three months: a multi-stage classification pipeline (~21k organisations through Claude tool-use), a rubric-based scoring system over ~6k candidates, a daily partner-facing intelligence pipeline, and the underlying Postgres/Supabase data infrastructure. Run the firm's AI rollout in parallel: 1:1 discovery sessions with every member of the investment team, workshop programme, AI block at the 2026 firm offsite.
Systems Modelling Lead
2022–2025Aurelle
Owned business systems across finance, supply chain, and fulfilment during 200%+ annual growth. Maintained 99.5% on-time delivery across 206,000+ orders. Built forecasting and procurement automation pipelines with explicit invariant enforcement.
Systems Engineering Associate — Regulated Financial Systems
2019–2022KPMG
Translated regulatory requirements (FDIC 370) into system and data-architecture modifications across core banking environments at Tier-1 financial institutions.
Systems Engineering Intern
2016–2019Siemens
Built automation tools, data models, and internal applications (iOS) for complex engineered product lines.
Research
Topology–Orientability–Chirality Framework for Neural Dynamics
A three-layer inference framework for classifying latent manifold structure in high-dimensional neural time series, with application to fNIRS neuroimaging data from surgical neuroergonomics studies. Designed persistent homology pipeline distinguishing toroidal from non-orientable manifolds via coefficient-field sensitivity. Introduced geometric coupling — a structural dependence concept detecting constraints between neural signals invisible to correlation or mutual information. Part of a broader Geometric Cognition Programme exploring topology, geometry, and dynamics as a unified language for biological and artificial intelligence.
Geometric Cognition Programme
A multi-phase research programme investigating cognition as structured dynamics on latent state spaces. The programme spans topological backbone detection, fiber bundle models of neural representations, geometric coupling between cognitive subsystems, and comparative geometry across biological and artificial intelligence. The accompanying coggeometry-studio provides interactive visualisations of the core mathematical concepts.
Selected Technical Papers
- Temporal Coherence in Long-Horizon Sensor Systemsinterval time, provenance, queryable uncertainty
- Latency and the Effective Capital Blueprintfeasibility sets, threshold exclusion
- Phase Legibility as Infrastructurecoordination friction, falsifiable tests
- Gradient Encoding in Human Oversight Systemsbounded risk encoding, boundary conditions
Projects
coggeometry-studio
Interactive geometric cognition explorer. Visualises manifold inference, persistent homology, orientability diagnostics, and geometric coupling concepts. Live demo →
ticker
Multi-agent LLM simulation platform. Deterministic agent-economy with escrow-gated settlement, structured tool use, and cross-model evaluation across Anthropic and OpenAI APIs.
cortex-rag
Document Q&A with retrieval-augmented generation. FastAPI, ChromaDB, Claude API. Multiple chunking strategies with retrieval quality evaluation (faithfulness, relevance, context precision).
kv-cache-tradeoffs
LLM internals analysis. KV cache compression and attention mechanism trade-offs with quantitative memory-quality analysis across degradation regimes.
Education
MSc Computer Science (AI/ML Focus)
2025–2026University of Birmingham
Dissertation: geometric inference for neural dynamics (supervised by Dr Felipe Orihuela-Espina). Machine Learning, Software Architecture, Data Systems, Algorithmic Reasoning.
BSc Systems Engineering
University of Georgia
Dean's List. Foundation in systems thinking, formal modelling, and engineering design.
Technical Skills
AI & LLM Systems
OpenAI API, Anthropic API, Hugging Face, RAG pipelines (ChromaDB, FAISS), structured generation, evaluation frameworks
ML & Research
PyTorch, persistent homology (Ripser, GUDHI), topological data analysis, time-series analysis, statistical inference
Languages & Core
Python (NumPy, SciPy, Pandas, Pydantic, asyncio), SQL, Git/GitHub
Infrastructure
FastAPI, Flask, SQLite, DuckDB, Docker, CI/CD pipelines
Writing
- Technical papersPDFs