Allan GaigherConsulting Data Engineer & Architect

Portfolio

A few projects that show how I work.

Details are anonymised but representative of the kind of problems I solve: migrations, reporting, and applied AI in financial and data-heavy environments.

Applied AI & internal assistants

LLM-powered tools that sit inside a secure perimeter and automate the boring, error-prone parts of data and reporting work.

Reconciliation assistant for a mid-market asset manager

LLM assistant that guides ops through daily reconciliation.

~3h → ~45min per day
Tech used
PythonLangChainAzure SQLPower BI
Click for more details

Operations teams spent hours each day stepping through spreadsheets and checks to reconcile positions and cash.

Built an internal LLM assistant that guides operations staff through daily reconciliation, using existing controls and playbooks.

  • Ingested trade, position, and cash data into a governed warehouse.
  • Wrapped existing reconciliation logic with a conversational interface.
  • Added explanations and audit trails for every suggested action.

Full audit trails for every suggested action.

Regulatory Q&A bot for policy documents

Retrieval-augmented assistant over internal policy docs.

1000+ documents indexed
Tech used
PythonOpenAIPineconeAzure
Click for more details

Analysts and compliance teams repeatedly asked the same policy questions, digging through PDFs and emails for answers.

Implemented a retrieval-backed assistant over internal policy documents to answer routine regulatory questions from analysts and compliance.

  • Built document ingestion and chunking pipelines with metadata tagging.
  • Deployed a retrieval layer with strict access control by team and region.
  • Logged and reviewed prompts/answers to tune relevance and safety.

Strict access control by team and region; prompt/answer logging for compliance.

Reporting & analytics (Power BI, Tableau)

Dashboards and semantic models that answer real questions for finance, risk, and operations — not just vanity metrics.

Executive performance dashboards for a PE-backed fintech

Board-ready Power BI model for recurring performance reporting.

Single trusted model
Tech used
Power BIAzure SynapseSQL Server
Click for more details

Leadership relied on inconsistent spreadsheets for ARR, churn, and unit economics.

Designed and delivered a Power BI model used by the exec team for weekly performance and board reporting.

  • Modelled ARR, churn, and unit economics from multiple source systems.
  • Implemented row-level security for finance, sales, and operations.
  • Documented KPI definitions so finance and product spoke the same language.

Row-level security for finance, sales, and operations.

Front-office analytics for a trading desk

Low-latency P&L and limits dashboards for traders and risk.

Intraday refreshes
Tech used
Power BIAzureSQLdbt
Click for more details

Traders and risk teams had fragmented views of P&L, exposure, and limits across tools.

Built low-latency dashboards to monitor daily P&L, risk, and limits with intraday refreshes.

  • Designed a warehouse model optimised for fast slice-and-dice queries.
  • Added alerting on rule breaches directly into the reporting layer.
  • Worked with desk heads to simplify layouts to what they actually used.

Alerting on rule breaches built into the reporting layer.

End-to-end data platforms & pipelines

Data platforms that collect, clean, and serve data reliably — often in regulated or audit-heavy environments.

Cloud migration of a legacy reporting stack

Migrated on-prem reporting stack to a modern cloud data platform.

Zero-downtime cutover
Tech used
AzureSQLSynapsePower BI
Click for more details

Critical reporting ran on ageing on-prem SQL and Excel workflows that were hard to change and fragile.

Led the migration from on-prem SQL and Excel workflows to a modern cloud data platform.

  • Designed the target architecture on Azure with managed services.
  • Ran dual-running period with automated reconciliation between old and new stacks.
  • Trained the internal team to own and extend the platform after go-live.

Dual-running with automated reconciliation before cutover.

Data platform for a multi-strategy fund

Central platform feeding risk, finance, and front-office reporting.

Shared data model
Tech used
PythondbtSnowflakeAirflow
Click for more details

Different teams maintained their own data extracts and definitions for core measures.

Helped design and build a central platform feeding risk, finance, and front-office reporting.

  • Implemented ingestion patterns for market, reference, and trade data.
  • Collaborated with risk and finance to align data definitions and controls.
  • Set up monitoring so teams saw issues before auditors did.

Monitoring and controls so teams see issues before auditors.