Specialist in applied artificial intelligence, experienced in designing and deploying production machine-learning systems for identity verification, due diligence, and financial-crime controls. My work spans end-to-end model development, AI application architecture, LLM-enabled agentic workflows, and client-facing advisory projects in regulated environments. It reflects strong technical depth and a product-minded approach to delivery, underpinned by academic rigour.
CORE EXPERTISE
- Applied AI for identity verification, anti-spoofing, forgery detection, name screening, and due diligence systems.
- AI application architecture, including Python-based HTTP services (e.g., Flask/Gunicorn, FastAPI/Uvicorn), event-driven systems (e.g., Kafka, RabbitMQ), and scalable, resilient deployment patterns.
- Deep learning, computer vision, NLP, and LLM-integrated systems, including embeddings, entity extraction, sentiment analysis, and agentic workflow design.
- Cloud infrastructure, DevOps, and MLOps using AWS, GCP, Docker, Kubernetes, and TorchServe.
- Technical leadership of small Agile R&D teams, including delivery planning, Jira- and GitHub-based workflows, and technical review.
- Programming and data tooling, with strong Python and SQL experience, plus additional experience in SAS and Fortran, and working familiarity with C++, Java, HTML, R, and MATLAB.
EXPERIENCE
Head of Artificial Intelligence Research, Idenfo, London (Hybrid), Jul. 2025 - Present
- Leading end-to-end research, design, and delivery of AI-enabled products and features for identity verification and due diligence, spanning problem definition, architecture design, data strategy, model development, evaluation, and production deployment.
- Managing a small R&D team with two direct reports, while also flexibly coordinating additional contributors when portions of their time are allocated to research and product-development initiatives.
- Improving name matching and public-domain screening systems through the integration of LLMs, text embeddings, entity extraction, sentiment analysis, and other NLP techniques; over the past 12-18 months, guiding the company's adoption of agentic LLM-powered systems and leading the delivery of a semi-automated agentic workflow for adverse media and politically exposed person checks.
- Building a production-ready face screening system around the company's proprietary facial-comparison model, including vector search and supporting infrastructure for deployment in real-world screening workflows.
- Integrating AI-assisted workflows within the development team, including evaluation of tools such as Cursor, Claude, and GitHub Copilot, engineering of system prompts, and establishment of governance rules for compliant use.
- Leading advisory projects for several Dubai-based banks through Idenfo, including on-site system architecture and data-quality reviews, as well as anti-money-laundering and anti-fraud tuning projects, now with responsibility for leading engagements, supervising Idenfo team members contributing to those projects, and assisting with the training of clients' junior staff.
Innovation Specialist - AI & Machine Learning, Idenfo, London (Hybrid), Feb. 2023 - Jul. 2025
- Took on increasing responsibility for the design and development of AI-enabled products and features across identity verification and due diligence, and played a key role in each stage of development.
- Developed a facial comparison model from scratch that achieved 99.7% accuracy, with a false-acceptance rate < 1 x 10-4%, on a large, diverse, and representative dataset; the model is now integrated into Idenfo's IDV platform and used to verify customer identities for a wide range of clients, including Pakistan's largest microfinance bank.
- Supervised junior staff on selected R&D tasks, reviewed technical work, and maintained high standards across development projects.
- Became the primary point of contact during selected advisory projects, progressing from a supporting analytical role to more direct responsibility for client communication and delivery.
Innovation Associate - AI & Machine Learning, Idenfo, London (Hybrid), Jul. 2022 - Feb. 2023
- Worked primarily as an individual contributor on applied AI research and development for identity verification and public-domain screening products.
- Focused on improving the company's public-domain name-screening product, and on researching anti-spoofing methods for upcoming identity-verification products.
- Contributed data analysis and technical support to several advisory projects involving anti-money-laundering and anti-fraud systems, including work on large financial datasets and transaction-monitoring rule analysis.
Mathematics and Science Tutor, Freelance, Frome, England, 2019
Intern, Centre for the Analysis of Motion, Entertainment Research and Applications, Bath, England, 2017
- I processed data from optical motion capture systems.
Assistant Database Administrator, Kensington Domestic Appliances, Heathfield, England, 2015 - 2016
- I managed the database supporting the online store.
PUBLICATIONS
Warsop, S.J.E.T., Caixeiro, S., Bischoff, M., Kursawe, J., Bruce, G.D., and Wijesinghe, P., Estimating full-field displacement in biological images using deep learning, npj Artif. Intell. 1, 6 (2025). https://doi.org/10.1038/s44387-025-00005-x:
The estimation of full-field displacement between biological image frames or in videos is important for quantitative analyses of motion, dynamics and biophysics. However, the often weak signals, poor biological contrast and many noise processes typical to microscopy make this a formidable challenge for many contemporary methods. Here, we present a deep-learning method, termed Displacement Estimation FOR Microscopy (DEFORM-Net), that outperforms traditional digital image correlation and optical flow methods, as well as recent learned approaches, offering simultaneous high accuracy, spatial sampling and speed. DEFORM-Net is experimentally unsupervised, relying on displacement simulation based on a random fractal Perlin-noise process and optimised training loss functions, without the need for experimental ground truth. We demonstrate its performance on real biological videos of beating neonatal mouse cardiomyocytes and pulsed contractions in Drosophila pupae, and in various microscopy modalities. We provide DEFORM-Net as open source, including inference in the ImageJ/FIJI platform, for rapid evaluation, which will empower new quantitative applications in biology and medicine.
EDUCATION
Master's in Physics (First-class, Integrated), University of St Andrews, Scotland, 2018 - 2022:
- Completed a research-based Master's project, "Measuring biomechanics with deep learning", which became the basis for the publication, "Estimating full-field displacement in biological images using deep learning", above.
- Studied a wide range of topics across Physics, Astronomy, and Mathematics, including:
- quantum field theory,
- Monte Carlo simulation,
- magnetohydrodynamics,
- electronics, and
- computational astrophysics,
- signal processing.
- Named on the Dean's List (for averaging first-class results across all modules) in every academic year.
- Awarded the Astronomy and Astrophysics Medal for the highest academic performance during Junior Honours.
- Awarded the Physics Medal for the highest academic performance during my final year.
- Awarded the MPhys Project Prize for the year's best Master's project.
A Levels, Frome Community College, England, 2016-18:
- Physics (A*), Mathematics (A*), Chemistry (A), and Photography (A*): the highest results in the year group.
