Suvro Ghosh

Bootstrapped and architected proprietary Health Information Exchange (HIE) and Clinical Trial Management System (CTMS) platforms, bridging the gap between legacy clinical data and Applied AI readiness.

Designed mobile-enabled clinical data tools aligned with CDISC standards, accelerating trial execution and ensuring strict regulatory compliance for Phase II/III trials.

Led full lifecycle development: system architecture → data modeling → validation logic → database lock, ensuring data integrity and analytical usability under conflicting constraints.

Operated as an independent architect partnering with healthcare clients to navigate interoperability challenges and translate clinical workflows into structured, scalable semantics.

Focused on cloud-native thinking in environments where legacy assumptions still dominate.

Explored SaaS architectures for complex systems where multi-tenancy introduces not just scaling concerns, but semantic isolation problems.

Worked on large-scale healthcare data warehouses integrating VA hospital data and national datasets. The challenge was not ingestion. It was reconciliation.

Flat files, SAS datasets, relational tables—each internally consistent, none aligned. Built pipelines that transformed data, knowing that every transformation introduces loss.

Performed statistical analysis across diabetes, ESRD, and nephropathy studies, where variable definition often mattered more than the analysis itself.

Built and maintained research databases sourced from MUMPS-based VA systems, flat files, and relational structures.

Designed ETL pipelines using SSIS where "integration" meant negotiating between incompatible representations.

Implemented outlier detection not as a statistical afterthought, but as a structural necessity due to inconsistencies in operational data.

Supported datasets spanning inpatient, outpatient, pharmacy, and laboratory domains, each with its own implicit ontology.

Migrated unstructured systems into relational SQL Server architectures.

Redesigned schemas, indexing strategies, and ETL workflows.

The technical work was straightforward. The challenge was translating assumptions embedded in legacy data into explicit structure.

Built reporting systems from mixed physical and digital data sources.

Designed SQL-based reporting pipelines, where the primary constraint was not tooling, but the absence of consistent data semantics.

Taught Operating Systems, Computer Networks, and Network Security.

Explaining systems forces clarity. Most production systems lack that clarity.

Implemented multi-queue caching algorithms for high-performance systems.

Early exposure to the idea that system behavior is shaped less by code and more by constraints and access patterns.