Projects

2026

Synthetic evaluation lab for AI-agent reliability: 350 golden cases, 60 red-team cases, RAG evaluation, safe refusal, safety classifiers, prevalence estimation, human-review simulation, mitigation impact, release gate reporting, FastAPI, OTel tracing, and CI.

2026

Reproducible benchmark measuring how published adversarial prompts perform against 2026-era LLMs and whether prompt-only defences move the needle — with cross-judge validation and bootstrap confidence intervals.

2026

Zero-shot event extraction with Qwen2.5-7B-Instruct on MAVEN and WikiEvents. Compares unconstrained vs constrained-label prompting across trigger detection, type prediction, and argument extraction. A100 GPU inference via Hugging Face.

2026

End-to-end synthetic fintech data product — dbt metrics, CUPED A/B experimentation, activation model, geo-lift referral analysis, pricing intelligence, FastAPI service, and a full GCP deployment path with BigQuery, Cloud Run, and Cloud Monitoring.

2026

Extractive document question-answering pipeline using PDF text extraction, sentence-based chunking, RoBERTa-SQuAD2 inference, and answer evaluation scripts.

2026

End-to-end analytics engineering on 4.99M UK Land Registry property transactions. dbt + DuckDB warehouse with staging, intermediate, fact, dimension, and reporting layers, 88 data tests, CI, docs, and a Streamlit dashboard.

2026

Distributed data mining and ML pipelines on 1.9M-20M record datasets, run on the University of Sheffield Stanage HPC cluster. Web log mining, traffic prediction, HIGGS classification, MovieLens recommendations.

2025

Classical ML benchmark on FBANK speech features. Improved speed-classification accuracy from 79.2% to 86.6% via feature standardisation and kNN tuning. Compared kNN, Logistic Regression, Linear SVM, Random Forest with detailed error analysis.