CZCi ZhuData & AI leader

Based in Greater Toronto Area, Canada

00

Ci Zhu / Data & AI leaderBuild systems and teams, Ship intelligence.

Enterprise scale. Engineering depth. About a decade connecting data, business decisions, and practical AI.

Explore my work ↘Start a conversation ↗
Ci Zhu
Profile / 01Ci ZhuData platforms + applied AI
018Departments empowered with data

Data strategy, service and field, sales, supply chain, finance, marketing, billing and collection, and builder operations.

0210–15%Lower data-platform costs

Azure SQL, compute, and Microsoft Fabric optimization.

0330%Shorter cycles. Stronger release quality.

AI-driven review enforces best practices. A standard workflow streamlines deployment requests through release.

0425+Systems and external platforms integrated

Career-wide: ERP, CRM, payments, third-party databases, APIs, and SFTP. Batch and streaming.

01 / SELECTED OPERATIONS

Outcomes over theatre.

Platforms modernized. Data connected. Decisions improved.

Platform modernizationCASE / 01

Enercare · Professional experience

Microsoft Fabric, in stages

Led platform design, rollout, and adoption. Kept reporting continuity, data access, and cost at the centre of the transition.

OUTCOME / SCOPE

10–15% lower data-platform costs across Azure SQL, compute, and Microsoft Fabric; approximately 600 legacy Power BI Pro licenses retired.

Microsoft FabricAzure SQLPower BILakehouseAzure DevOps
Read the full case

The challenge

Modernize the platform while preserving reporting continuity, expanding data access, and controlling cost.

My role

Led assessment, architecture, sequencing, and business adoption. The team implemented and validated the workloads.

The approach

Started with parallel Power BI workspaces and gradual cutover. Added Azure SQL mirroring and Lakehouse shortcuts, then moved reporting workloads and tuned performance.

Later ingestion and near-real-time work remained a separate phase. Training, communications, and fallback plans supported each transition.

The result

Reported 10–15% lower data-platform costs across Azure SQL, compute, and Microsoft Fabric. Approximately 600 legacy Power BI Pro licenses retired.

02 / EXPERIENCE LOG

From query to strategy.

Engineering foundations. Enterprise responsibility.

1
2024–Present

Enercare

Senior Manager, Data Strategy and Analytics

Leading enterprise data strategy, Fabric modernization, and applied AI. Delivering data services across eight departments.

Responsibilities and scope

4 full-time engineers, 2 contractors, and 2 dotted-line Power BI developers.

Built a shared delivery team spanning engineers, contractors, and dotted-line Power BI developers.

Led platform design, staged migration, and business adoption. The team implemented and validated the workloads.

Delivered retrieval-based AI, reusable agent skills, and Azure DevOps pipelines. Reported results: 10–15% lower platform costs across Azure SQL, compute, and Fabric; approximately 600 legacy Pro licenses retired; 30% shorter deployment cycles with AI-driven review and a standard release process.

2
2017–2023

LG Electronics

Business Intelligence Developer → Senior Manager

Four roles. Growing responsibility for data architecture, analytics, and people leadership.

2022–2023

Senior Manager, Digital Transformation and Data Science

Led forecasting, commerce, marketing, retail-channel, and financial analytics.

2021–2022

Manager, Digital Transformation and Data Science

Led the analytics team and aligned the roadmap with business and IT leaders.

2018–2021

Business Intelligence Architect

Designed data models and integrations. Led BI for the employee and partner store.

2017–2018

Business Intelligence Developer

Built SQL, ETL, and reporting foundations.

Responsibilities and scope

Led three direct developer reports and four dotted-line departmental analysts.

Integrated 20+ systems and platforms at LG, processed 100+ files daily, and automated reporting that recovered approximately 60 hours a week across six Product Managers.

Recognition: LG Excellence Award · LG Ovation Award

PEOPLE & CHANGE

How I lead

Clear direction. Strong teams. Lasting change.

Align the ambition

Connect business priorities, platform strategy, and investment decisions.

Build strong teams

Give people clear ownership, room to grow, and a shared standard of excellence.

Make change stick

Turn strategy into adoption through trust, accountability, and disciplined delivery.

03 / AI ENGINEERING

Engineer the agent,
not the demo.

Useful applications. Reusable expertise. Inspectable results.

A–01CAPABILITY

AI in the workflow

Start with the business task.

Led retrieval-based knowledge applications with Azure AI Foundry and AI Search for customer-service work.

Retrieval-based knowledgeCustomer-service workflowsAzure AI
A–02CAPABILITY

Reusable agent skills

Turn expertise into repeatable workflows.

Versioned skills for data catalogs, relationship mapping, and transformation logic—with explicit inputs and reviewable outputs.

Agent skillsData engineeringReusable workflows
A–03CAPABILITY

Coordinated development

Separate the work. Strengthen the review.

Focused roles for exploration, implementation, and independent review. Clear handoffs and maintainable context.

Multi-agent workflowsReviewMaintainability
A–04CAPABILITY

Evidence and control

Keep the limits visible.

Preserve sources, distinguish suggestions from approved definitions, and label synthetic data clearly.

SourcesHuman reviewClear boundaries
PUBLIC PROJECTS

Engineering in the open.

Knowledge. KPI governance. Better engineering workflows.

Open-source toolkit

MirrorArc

Technical alpha

Public repository and educational demo.

Knowledge with a source trail.

Connect original files, Markdown, and working knowledge. See where information came from, how it connects, and when it needs refreshing.

Technical alpha. The Ontario electricity workspace is educational: no live grid data, forecasting, alerts, or energy-sector affiliation.

Explore the demo ↗Read the repository ↗Project overview ↗
Problem, contribution, and approach
The problem

Sources, summaries, and decisions lose context when their relationships disappear.

My contribution

Architecture, workflow design, and implementation with AI-assisted development.

  • Keep original records authoritative.
  • Expose source relationships and lifecycle state.
  • Connect catalog, metadata, and document views.

Make existing knowledge easier to inspect and maintain.

PythonMarkdownSource mirroringKnowledge relationships
Public engineering project

MetricGraph

Working synthetic-data demo

Synthetic demo. Enterprise connectivity in development.

Know what a KPI means.

Connect KPI definitions to calculations, owners, versions, and usage. Make conflicting measures and change impact easier to understand.

Synthetic metadata and browser-session interactions only. No validated live Fabric tenant connection or live DAX execution.

Explore the synthetic demo ↗Read the repository ↗
Problem, contribution, and approach
The problem

Reports multiply faster than agreement on what the numbers mean.

My contribution

Domain modeling, architecture, interface design, and implementation.

  • Separate business definitions from technical measures.
  • Track versions, variants, ownership, and approvals.
  • Show relationships and change impact.

Start with a business question. Make the definition traceable.

Microsoft Fabric / Power BI metadataReactTypeScriptRelational registryGraph exploration
Open-source reference toolkit

Vibe Coding Repository Standard

Public preview

Codex-first reference implementation.

Keep AI-assisted code understandable.

A reference toolkit for focused agent instructions, current documentation, and changes people can review.

Independent community project. No vendor endorsement, official-standard status, or validated universal agent compatibility.

Explore the toolkit ↗
Problem, contribution, and approach
The problem

Fast code generation can outpace shared understanding.

My contribution

Workflow design, repository practices, and reference implementation.

  • Inventory the repository before restructuring.
  • Keep permanent instructions small.
  • Separate implementation and review.

Faster generation should preserve clear understanding.

Repository standardsAgent workflowsPython validatorDocumentation practices
04 / EDUCATION

Theory with
operating range.

Mathematics, intelligence, and business judgment.

THE THROUGHLINE

Rigor meets application.

Mathematical optimization, statistics, and AI management.

ACADEMIC RECORD / 012020–2021
Smith School of Business at Queen’s University

Smith School of Business at Queen’s University

Master of Management in Artificial Intelligence

Artificial intelligence and management

AI expertise connected to management decisions.

ACADEMIC RECORD / 022011–2015
University of Waterloo

University of Waterloo

Honours Mathematics — Mathematical Optimization

Operations Research Specialization; Joint Honours Statistics

Mathematical rigor for complex decisions.

05 / CURRENT WORK

In progress.

Current project states, at a glance.

Q—001TECHNICAL ALPHA

MirrorArc

Source-preserving knowledge for people and agents.

Q—002SYNTHETIC DEMO

MetricGraph

KPI discovery and governance with synthetic data.

Q—003PUBLIC PREVIEW

Vibe Coding Repository Standard

Focused instructions and reviewable AI-assisted code.

FIELD NOTES / LATESTView all notes ↗
N—001Governed knowledge beats another chatbot2026.074 min↗N—002A Fabric migration is an operating-model change2026.073 min↗N—003From AI demo to enterprise system2026.073 min↗
OPEN CHANNEL

Data strategy. Platform modernization. Practical AI.

Let’s turn complexity into a system that works.

Start a conversation ↗Follow the build ↗