BRYAN AGAS / BUSINESS INTELLIGENCE & AI

Data, AI &
the decisions
between them.

I turn complex data into useful reporting, build practical tools, and explore how AI can improve the way work gets done.

Explore selected work Connect on LinkedIn

HR DATA & INSIGHTS ANALYST · UK

THE WORK, CONNECTED01 — 04

Ask the right question.
Build what makes it actionable.Explore the connections

POWER BI & DAXSQL & DATA MODELLINGPYTHON TOOLSAGENTIC WORKFLOWSBUSINESS CONTEXT

01 / PROFILE

Close to the data.
Focused on the decision.

My foundation is business intelligence. My direction is the intersection of data, AI and business transformation.

I’ve progressed from data technician to Power BI development and data analysis, and now HR data and insights. That work connects data preparation, modelling and report design with the questions people need to answer.

Alongside professional reporting, I build Python tools and experiment with agentic BI workflows. The common thread is practical: understand the problem, make the logic clear, and keep a person accountable for the result.

Follow the experience

02 / SELECTED WORK

Useful systems.
Different scales.

From professional reporting to tools and experiments.
Each project starts with a specific problem.

REPORTING / ARCHITECTUREILLUSTRATIVE

SQLAWSEXCEL

SHARED DATA MODELOne definition. Consistent measures.

OPERATIONSHRCOMMERCIAL

A clearer path from source to decision.

01 / BUSINESS INTELLIGENCEProfessional work

Making reporting
work for the business.

Spreadsheet reporting creates repeated preparation and competing definitions. At Ipsum Utilities, my work spanned Power BI dashboards for operational, HR and commercial needs.

My contribution: Data preparation, SQL, modelling, DAX and report design, shaped with stakeholders.
Approach: Reusable datasets and shared measures, with reports organised around business questions.
Result: Delivered reporting across business functions. Employer dashboards and data remain private.

Power BI
SQL
DAX
AWS

Read the delivery approach

Commercial analysis: rebuild legacy Excel reporting for quotes and commercial questions with Power BI and SQL.

Workforce reporting: model headcount, turnover and demographics with attention to interpretation and access.

Delivery: clarify the decision with stakeholders, develop the model, check measures and iterate on presentation. No internal screenshots or unverified savings figures are published.

DATA DICTIONARY / SEARCH

Find a column definition

COLUMN → DEFINITION

The right context,
without the spreadsheet hunt.

ILLUSTRATIVE INTERFACE · NO EMPLOYER DATA

02 / DATA TOOLINGSource inspected

Make the meaning
easy to find.

A data dictionary is only useful if people can find the definition they need. I built a desktop search tool over an Excel dictionary using Python, pandas and Tkinter.

My role: Tool design and implementation.
Outcome: Column lookup and matching definitions in a dedicated interface; a separate browser version explores the same task.
What it demonstrates: Practical problem selection, data handling and interface design. No usage or time-saving metric is claimed.

Python
pandas
Tkinter
JavaScript

Read the technical scope

The Python implementation reads an Excel sheet into pandas, searches column names and displays matching definitions using Tkinter. The browser implementation is a separate HTML/JavaScript tool.

The source was inspected for this portfolio. The workbook and source repository are not linked publicly to protect potentially sensitive metadata.

AGENTIC BI / WORKFLOW DESIGN

Business question

ProfileModelBuild

Human review & QA

DESIGN EXPLORATION · NOT A PRODUCTION SERVICE

03 / PRACTICAL AIExperimental

Design the handoffs.
Then add the agents.

I’m exploring how a BI delivery process can be divided into specialist agent roles, with explicit inputs, outputs and review gates.

My role: Workflow design, role definition and experimentation.
Approach: Separate profiling, modelling, build and QA; define ownership and bound revision loops.
Current state: Experimental design, alongside a Python autonomous BI prototype. Production reliability and business impact have not been independently validated.

Agent orchestration
Python
Quality gates

Read the experiment

The workflow explores specialist roles across discovery, profiling, modelling, DAX, report design, security, QA and documentation. Explicit handoffs make each stage reviewable.

A related Python prototype explores a staged path from profiling and transformation to KPIs, anomalies and an HTML dashboard. It remains a prototype; an automated result still needs validation against trusted data.

ALSO IN PUBLIC CODE A small Python utility for reading GitHub activity.

Inspect the source

03 / CAPABILITIES

The craft behind
the outcome.

Technical work with business context.
Clear boundaries around emerging skills.

Business intelligence

Power BI, DAX, Power Query, SQL, star schemas and reusable datasets. Build the model before the visual.

Data & reporting

Operational, commercial and HR analytics. Data quality, consistent definitions and reporting that supports a decision.

AI & automation

Python tools, staged analytical prototypes and agent workflow design. Review, ownership and limits are part of the design.

Delivery & collaboration

Translate stakeholder needs into analytical requirements, iterate on reports and mentor junior analysts in Power BI and DAX.

04 / PROFESSIONAL EXPERIENCE

A growing scope.
A consistent thread.

From preparing the data to shaping how it is used.

View professional profile

2026 — PRESENT

HR Data and Insights Analyst

Addleshaw Goddard

Applying a BI background to people data, reporting and insight. Clear interpretation and appropriate access are central to the work.

SEPTEMBER 2023 — 2026

Power BI Developer & Data Analyst

Ipsum Utilities

End-to-end reporting across operational, HR and commercial needs: data preparation, models, measures and stakeholder-facing dashboards.

JULY 2022 — SEPTEMBER 2023

Data Technician

Ipsum Utilities

Excel KPI dashboards, data migration support and quality assurance, building a foundation close to the source data.

05 / HOW I THINK

Start with the decision.
Earn the automation.

The same discipline connects reporting, tools and AI: make the work understandable before making it faster.

01
Find the actual question.
The report is a means to a decision. Agree who needs to act, what they need to know and when.
02
Make the meaning explicit.
Shared definitions in BI and searchable metadata in a data dictionary solve the same problem: ambiguity.
03
Keep the result reviewable.
Agent roles need clear handoffs. Analytical outputs need checks. A person remains responsible for consequential decisions.

06 / CURRENT EXPLORATION

Building towards
more useful AI.

ONGOING / OCTOBER 2026

AI-ready business intelligence

Exploring semantic descriptions, business glossaries and known-answer questions so AI has a clearer analytical foundation.

RESEARCH & PROTOTYPING

Governed enterprise workflows

Writing about Workday, people data and AI accountability: access boundaries, evidence, exception handling and service ownership.

Read my thinking on LinkedIn

07 / THE NEXT CONVERSATION

Good data is a start.
What should it change?

I’m interested in conversations about senior BI and analytics opportunities, practical AI adoption, and data-led transformation.

Discuss an opportunity Explore public GitHub

Recruiters: send the role, scope and team context via LinkedIn. A current CV can be requested there.

THE COMBINATION

Business context.
Analytical depth.
A builder’s instinct.

Explore the interactive portfolio

Data, AI &the decisionsbetween them.

Close to the data.Focused on the decision.

Useful systems.Different scales.

Making reportingwork for the business.

Make the meaningeasy to find.

Design the handoffs.Then add the agents.

The craft behindthe outcome.