Ivan Damnjanovic | Principal Data Platform Architect

About

I am a Principal Data Platform Architect building governed lakehouse platforms where contracts, lineage, and entitlements define the product before any consumer arrives. Unity Catalog governance, ODCS YAML contracts, ISO 8000 quality frameworks, and BCBS 239 lineage with compliance-ready evidence are design constraints from Bronze onward, not post-incident add-ons. Dashboard-first stacks can look fine in a demo. They are harder to defend when audit asks for catalog-native evidence.

I founded and operate Ambient Systems as principal architect of its production governed data platform and contributed to Ambient Core open patterns so model and agent paths route through contracts and catalog. Background spans Deloitte analytics, University incubator data governance, solar energy, college teaching in AI, Operating Systems (Linux), and entrepreneurship, plus doctoral research in business management. The same systems-boundary discipline applies whether the consumer is a regulator, a BI tool, or an agent runtime.

Ivan Damnjanovic - Principal Data Platform Architect

Work

Selected delivery examples by challenge, capability, delivery, and output
Challenge	Capability	Delivery	Output
PlatformOperators needed auditable Gold for BI, sharing, and compliance evidence. Ungoverned warehouses with dashboards bolted on later appear reasonable under delivery pressure, but they leave lineage and entitlements thin when scrutiny arrives.	Own the platform boundary end to end so every consumer path stays contract-backed rather than ad hoc.	Multi-tenant medallion lakehouse on Databricks and Unity Catalog; ODCS YAML contracts for schemas, SLAs, quality, and lineage; ISO 8000-aligned scoring; BCBS 239 retention; PySpark and Delta Lake into governed adapters with RLS and column masking.	Production Gold for Power BI, Tableau, and Delta Sharing through the catalog. Consumption adapters stay thin; correctness stays in the product contract.
Program governanceA $25M technology incubator supporting more than 2,000 organizations needed reliable data flows and reporting under audit expectations. Policy binders alone do not survive review when operators cannot see stewardship on the path.	Make program-scale data governance operational rather than documentary.	Structured data flows, reporting, and controls so stewardship and lineage stayed visible to operators and reviewers under university and program constraints.	Audit-ready program reporting at portfolio scale. Governance became a property of the delivery path, not a separate policy binder.
Solar operationsAn energy operator needed scalable data products and automation tied to the operating business. Models that live only in decks may look complete, but they do not move revenue or cost by themselves.	Couple product definition to the data path that runs solar operations.	Built and steered operational data products with automation across commercial and cost levers, keeping measurement close to the operating path.	Roughly 2x revenue growth and roughly 40 percent lower operational cost where those metrics were measured. Operating results follow the data path, not the slide deck.
Industrial analyticsFinancial-services and energy clients needed analytics that changed cost outcomes across a multi-engagement portfolio (nineteen engagements). Slide-only recommendations are easy to socialize and hard to bank.	Deliver analytics models that moved energy cost for an industrial mining client, not recommendations that end at the deck.	Analytics and modeling grounded in operational energy data under consulting delivery constraints and engagement sequencing.	Roughly 12 percent lower energy cost for the mining engagement. Measurable industrial savings still need a contract between model and operating data.
Open patternsTeams shipping intelligence into production risked opaque integrations that bypass governance and catalog. Local optimization here may appear reasonable, but it is often difficult to audit later.	Keep model and agent paths routed through governed contracts and catalog rather than ad hoc mappings.	Published open patterns for governed data products, orchestration, and contract-backed Gold outputs so the boundary is inspectable outside any single private stack.	A reusable boundary: intelligence consumes through catalog-native products. Under current conditions, the same principle holds in open patterns and in production platforms.