Selected projects and platform contributions

Problem

How can heterogeneous data sources be ingested, indexed, and queried without forcing every workload into a single storage model?

Approach

Built polystore ingestion and planning methods across databases, graph stores, text indexes, and analytical engines.

Contributions

• Designed heterogeneous ingestion and integration workflows.
• Developed query-processing methods for cross-store access.
• Contributed to patentable database-system ideas.
• Translated research concepts into platform-oriented system design.

Outcomes

• Enabled published work on polystore ingestion and query planning.
• Connected to patentable ingestion and query-processing methods.

Problem

How can a national data platform make distributed scientific datasets discoverable without flattening their semantics or provenance?

Approach

Built federated-search services around metadata normalization, text and semantic retrieval, APIs, and policy-aware access.

Contributions

• Built or shaped search-facing data models for scientific discovery.
• Worked on metadata and semantic access patterns.
• Helped design APIs for community use.

Outcomes

• Connected distributed scientific catalogs to a reusable discovery layer.
• Aligned with the National Data Platform's AI-ready data and workspace model.

Problem

How do wearable and clinical data streams get structured so they can support reproducible analysis without overfitting the platform to one study?

Approach

Supported continuous wearable-sensor ingestion, harmonization, monitoring, data-quality control, and analytical workflows.

Contributions

• Supported scalable ingestion for wearable and multimodal data.
• Contributed to analytics-oriented data organization.
• Helped enable predictive-health research workflows.

Outcomes

• Enabled a published conference paper and a large-scale wearable-sensing study.
• Connected platform work to symptom-prediction and monitoring workflows.

Problem

How can materials-science data remain discoverable and reusable across projects, workflows, and collaborators?

Approach

Built a collaborative data and analysis platform with metadata storage, FAIR search/access, JupyterHub analysis, and workflow logging.

Contributions

• Supported collaborative metadata design.
• Worked on research data discovery flows.
• Helped shape scientific workflow support.

Outcomes

• Published as a conference paper on collaborative quantum material science data management.
• Connected to a reusable quantum materials discovery and analysis workflow.

Problem

How can people ask complex public-health questions without requiring deep technical knowledge of the underlying systems?

Approach

Connected knowledge-based conversational engines to heterogeneous data integration and privacy-aware access patterns.

Contributions

• Supported secure data-service design.
• Contributed to knowledge-graph thinking for discovery.
• Helped make multimodal health data easier to explore.

Outcomes

• Aligned with the lab's AI-enabled query processing and decision-support direction.

Problem

How should an early-stage cloud startup build a usable platform before requirements and market expectations are fully stable?

Approach

Build the product and engineering foundation with enough rigor to survive changing requirements while keeping delivery practical.

Contributions

• Helped build the multi-cloud management platform.
• Contributed to product development in a startup setting.
• Helped establish the India engineering operation.

Outcomes

• Established the bridge between startup product delivery and operational engineering.
• Supported a cloud-management product across deployment, provisioning, monitoring, and releases.