Senior-led cloud · data · AI consulting
Cloudfish is led by Hubert Chan — a management consultant and principal solution architect who works from strategy through execution across multi-cloud, data and AI. Board-level framing, architecture that survives contact with delivery, and engineering that ships.
From boardroom framing to production operations
Engagement scope
On a North American retail cloud modernization, Hubert led data conversion delivery. The figures below describe the scope of that engagement — the environment and the migration — not attributed business outcomes.
The environment supported three retail banners totalling $9B in annual revenue. Role: led data conversion delivery within a broader modernization program.
Services
Bring a platform decision, a stalled migration, a governance gap or an AI idea that needs a real production path. The same person who frames it stays close enough to build it.
Framing for leaders who have to fund, defend and sequence the work: current-state assessment, target architecture, platform selection, cost posture and a roadmap that matches the organization's real operating model.
Target-state and solution architecture across cloud, data and integration — with the trade-offs, non-functional requirements and transition states written down before the expensive build work starts.
Data centre exits, warehouse and lakehouse migrations, and greenfield platform builds on Snowflake, Databricks, AWS and Azure — phased, reconciled and handed over to operations.
Master and reference data, lineage, cataloguing, quality controls, access models and the operating routines that keep trusted data trusted after the project ends.
Use-case selection, retrieval-augmented architecture, evaluation and guardrails — plus the harder part: turning a promising pilot into something a business can operate and support.
Interim platform or data leadership, delivery recovery, and cross-functional and multi-vendor coordination when a program needs senior hands rather than another deck — including programs with 50+ contributors and multiple delivery vendors.
Selected engagements
These engagements were delivered across Hubert's consulting, client-side and in-house roles. They describe his personal experience and responsibilities — not work delivered by Cloudfish Consulting as a firm.
01Retail & e-commerce · North America
Cloud modernization for a multi-banner retail group, exiting three data centres onto AWS and Snowflake. Led data conversion delivery: 260TB moved, 30+ databases, ~3,000 pipelines, ~4,000 reports and 20+ applications in 8 months, coordinating 50+ contributors across 4 vendors. The work continued past cutover into platform ownership, optimization and day-to-day operations.
02Automotive & manufacturing
Data conversion at petabyte scale for an SAP program spanning 30+ manufacturing plants, with 700+ source data feeds per plant to reconcile. Related work covered Customer 360 and master data management, including natural-language processing applied to messy operational and customer data.
03SaaS & technology
Led a team of 7 building an Azure data platform for a SaaS provider: Azure Data Factory and Airflow orchestration, dbt transformations, services on Kubernetes and Azure SQL as the serving layer, feeding near-real-time Tableau reporting for internal and customer audiences.
04Applied AI
Productization work on generative AI using AWS Bedrock, retrieval-augmented generation, LangChain and OpenAI models. Included customer journey and persona modelling, and executive-sponsored pilots used to establish where a generative approach would hold up in production — and where it would not.
05Tech startups & high growth
In-house and consulting roles inside startups and high-growth product companies — shaping architecture, data platforms and delivery as the product and the team grew. That work spans national real estate tech; green industry SaaS CRMs; travel & loyalty apps; and e-commerce shops.
Industries
Sector context changes the architecture. These are the environments where the constraints — regulatory, operational and commercial — are already familiar.
Reference experience
Organizations represented across Hubert's consulting, client and in-house experience. No endorsement or current affiliation is implied.
The principal
Cloudfish is deliberately senior-led. Engagements are shaped and carried by someone who has sat on both sides of the table — as a consultant, as a client and in-house as a platform owner.
Credentials
Credentials are a floor, not a differentiator — but at architecture level they keep vendor conversations honest. A representative selection is shown below.
Representative certifications
AWS Certified Solutions Architect – Professional
Microsoft Certified: Azure Solutions Architect Expert
SnowPro Advanced: Architect
Databricks Certified Data Engineer Professional
Google Cloud Professional Cloud Architect
Project Management Professional (PMP)Technology ecosystem
Technology is chosen to fit the operating model, not the other way around. The path from source system to decision is the thing being designed.
Snowflake · Databricks · Delta Lake · Apache Spark
AWS · Microsoft Azure · Google Cloud · Cloudflare
dbt · Python · SQL · Airflow · Azure Data Factory
SAP · ERP · CRM · APIs · event streaming
Tableau · Power BI · semantic layers · governed metrics
AWS Bedrock · LangChain · RAG · vector search · evaluation
Kubernetes · Terraform · CI/CD · secrets & identity
Data quality · lineage · cataloguing · privacy · cost management
How engagements run
Every engagement is scoped so there is something decision-ready early, and so your team owns what remains when it ends.
Clarify the business decision, constraints and measures of success.
Make architecture and delivery choices explicit before expensive build work.
Ship in thin, testable slices with quality, security and operations included.
Document, train and pair so your team owns what comes next.
Typical shapes
Start a conversation
Share the outcome, the current environment and where the work is stuck. You will get useful questions and a practical next step back — from Hubert, not a queue.
Good starting points
Prefer to talk first? Book a 30-minute call (opens in a new tab).