Profile
ML engineer with 9 years' total experience, the last two focused on
production LLM systems. Background in quantitative ML for financial
services — risk models, anomaly detection — which gave me rigorous
habits around validation, monitoring, and handling adversarial inputs.
Deliberately pivoting toward AI product engineering at SaaS companies
where I can see the end-user impact more directly. Strong Python, AWS,
and a growing depth in LLM orchestration frameworks.
Experience
Veritas Financial Technology · FinTech SaaS, 300 people, London
-
Owns Veritas's LLM inference infrastructure: model routing, prompt
caching, cost tracking, and failover across OpenAI and Anthropic
APIs. System handles ~180,000 completions/day at 99.7% uptime.
-
Built an AI-powered document analysis feature (regulatory filings,
audit reports) using a multi-stage pipeline: OCR → chunking →
embedding → retrieval → synthesis. Reduced analyst review time by
~4 hours per report.
-
Designed the evaluation framework: automated factual consistency
checks using an LLM judge model, cross-referenced against
ground-truth annotations. Enabled the team to ship two major
prompt versions per month with confidence.
-
Led the migration from a monolithic GPT-4 call to a tiered model
strategy (GPT-4o-mini for classification, GPT-4o for synthesis),
cutting inference spend by 41%.
-
Introduced prompt versioning and A/B infrastructure, enabling
controlled rollouts to subsets of the user base.
Veritas Financial Technology · same organisation, promoted
-
Built and maintained credit risk ML models (gradient boosting,
survival analysis) deployed into the Veritas underwriting
platform.
-
Developed the internal ML platform tooling: experiment tracking
(MLflow), feature store (Feast), and model registry. Reduced model
deployment cycle from ~3 weeks to 4 days.
-
Implemented real-time transaction anomaly detection (Isolation
Forest + LSTM) reducing false positive rate by 28% versus the
prior rule-based system.
Huxley Associates · Financial services consultancy, London
-
Delivered ML projects for tier-1 banking clients across credit,
fraud, and customer analytics domains.
-
Frequently worked embedded in client data science teams;
comfortable operating in regulated environments with formal model
governance requirements.
Barclays Corporate Banking · London
-
Rotational grad scheme across risk analytics and structured
products. Developed portfolio stress-testing models in Python and
SQL.
Skills
LLM / GenAI
RAG, prompt engineering, LLM evaluation, model routing, LangChain,
OpenAI API, Anthropic API, Chroma, pgvector
ML / Python
Scikit-learn, XGBoost, LightGBM, PyTorch (working knowledge),
FastAPI, Pydantic, Pandas
Infrastructure
AWS (ECS, Lambda, SQS, S3, RDS), Docker, Terraform, GitHub Actions,
MLflow
Data
PostgreSQL, Snowflake, dbt, Kafka
Education
BSc Mathematics and Computer Science (2:1)
University of Bristol
2011 – 2015
Back to the shortlistFICTIONAL SAMPLE CV