AI & GenAI Engineering Experience

Data Analyst / Engineer — GenAI-Powered Legacy Migration

A large-scale migration of on-premise data into a new Snowflake Enterprise Data Warehouse required converting a huge volume of legacy SAS scripts into SQL — a slow, error-prone task by hand.

• Built a GenAI workflow using Anthropic's Claude 3.7 to convert legacy SAS scripts into Snowflake SQL at scale
• Owned PII-masking across hundreds of SAS files, designing a Python Regex–based solution to identify and mask/map sensitive data
• Supported the ETL team writing SQL stored procedures across 3 staging layers and 1 target layer to migrate data into production

Business Analyst — Generative AI Risk & Compliance MVP

16,000+ compliance incidents are reported annually at this bank; 75% require a compliance assessment taking 2.5–5 hours to complete by hand. The client needed a GenAI copilot to cut that time down.

• Gathered requirements and translated them into delivery, bridging gaps between data scientists, solution architects, and Risk & Compliance business stakeholders
• Ran end-to-end testing and drove sign-off on production-gating artefacts (Test Plan, Test Approach, Detailed Technical Design, Test Completion Report)
• Ran daily standups and scrum ceremonies — backlog grooming, RAID tracking — to keep the build on schedule

Gen AI Engineer — Policy Simplification POC

The bank needed to summarise long, dense policy documents without losing their original intent or meaning — a task where getting the fidelity/compression trade-off wrong is costly.

• Deployed LLM API endpoints in Azure ML Studio to summarise lengthy policy documents
• Built a chunking pipeline to break large PDF documents into inputs the endpoint could consume
• Designed evaluation metrics and acceptance criteria to benchmark GPT-4o, GPT-4, GPT-3.5, and Llama 3 on summarisation quality and length reduction

Data Strategy Consultant — ML & MLOps Advisory

The client's Data Science team had built predictive ML models but needed help improving, communicating, and deriving ongoing value from them.

• Evaluated the client's existing predictive ML modelling and MLOps processes
• Outlined recommendations to streamline modelling and MLOps workflows in Python
• Delivered a roadmap for more effective stakeholder communication and migration away from legacy systems