Uma Chandar

Visionary Data &

Mission Systems Architect with 20+ years of experience supporting Federal, Postal Service, and National Security-adjacent programs, translating complex mission objectives into integrated technical architectures. Five-plus years supporting Federal programs (USPS, USPTO, Department of Veterans Affairs) via Accenture Federal Services, spanning enterprise architecture, System-of-Systems integration, technology modernization, and digital transformation initiatives. Proven success developing executive-quality architecture diagrams, technical documentation, and data flow/system context diagrams that connect operational needs to system capabilities. Extensive hands-on experience across data engineering, database administration, data modeling, AI/ML, data quality, and data governance across cloud platforms like Azure, AWS, and GCP. Expert in designing and scaling data lake/lakehouse environments (Snowflake, Redshift, Databricks) and implementing secure, real-time pipelines using Informatica, SSIS, Kafka, and Python. Known for driving innovation, efficiency, and stakeholder alignment across multidisciplinary technical and non-technical teams.

Gen AI Architect / Enterprise Solution Data Architect - Accenture Federal Services

(2022-01)

United States Postal Services (USPS) – Gen AI Architect / Enterprise Solution Data Architect

• Designed a scalable and modular architecture focused on future-proofing cognitive services, enabling integration with advanced AI and ML tools.
• Architected a Retrieve-and-Generate (RAG) architecture leveraging both information retrieval and LLM-based generative systems.
• Architected a system using LangChain and LLMs for document chunking, vector creation, and semantic retrieval.
• Implemented Databricks Lakehouse as the core RAG processing engine, using Delta Lake for version-controlled document chunks, embeddings, and metadata.
• Developed production-grade pipelines using Databricks Auto Loader to incrementally ingest and normalize documents with schema evolution and cloud-optimized file discovery.
• Leveraged Databricks Vector Search and Mosaic AI model serving for real-time semantic retrieval and low-latency LLM inference.
• Configured Databricks Photon-optimized Delta pipelines to accelerate embedding generation, dramatically improving compute efficiency.
• Used Databricks MLflow for model lifecycle management, tracking LLM fine-tuning experiments, evaluation metrics, and deploying models to serving endpoints.
• Implemented Unity Catalog (UC) for cross-workspace governance, secure credential passthrough, lineage tracking, and centralized access policy management.
• Integrated Azure AI Search and custom routing logic for real-time retrieval of document chunks, ensuring high accuracy in complex user queries.
• Designed the General Foundation Architecture for DaiSE Cognitive Services, enabling secure ingestion, scalable compute, and governed AI-powered data flows.
• Maintained architectural traceability from mission needs through technical implementation, documenting system interfaces, interoperability requirements, and integration points across the enterprise system landscape for stakeholder and IT leadership review.

Enterprise Solution Data Architect / Technology Leader - Accenture Federal Services

(2022-01)

USPTO – Enterprise Solution Data Architect / Technology Leader

• Architected scalable data pipelines and AI/ML systems using AWS, Azure, and GCP.
• Led Databricks Lakehouse modernization, consolidating siloed ETL, ML, and reporting workloads into Delta Lake–backed structures.
• Developed Delta Live Tables (DLT) pipelines implementing expectations, optimized data quality checks, and end-to-end lineage.
• Used Azure Databricks Repos with Git integration to standardize collaborative engineering workflows across data engineering teams.
• Created feature engineering pipelines using Databricks Feature Store, enabling consistent, reusable features across ML models.
• Increased processing performance by 170% by refactoring SQL workloads into Databricks SQL (DBSQL) warehouses running on Photon.
• Reduced operational cost by 45% via cluster right-sizing, cluster policy enforcement, autoscaling, and job orchestration optimization.
• Migrated legacy ETL jobs to Databricks Jobs with retry logic, task orchestration, and notebook-based execution.
• Implemented enterprise metadata governance using Unity Catalog + Lakehouse Federation to unify access policies across disparate data sources (AWS Glue, Azure SQL, Oracle).
• Designed NLP solutions including sentiment analysis, NER, and text classification, integrated with MLflow tracking and model registry.
• Collaborated across teams to deliver ML models deployed through Databricks Model Serving.
• Coordinated leadership reviews on pipeline architecture and created a best-practices handbook including Delta optimization (Z-Ordering, file compaction, OPTIMIZE, VACUUM) techniques.
• Fostered enterprise-wide education on Databricks compute tuning, Photon performance benefits, and DBSQL cost/performance trade-offs.
• Performed technical analyses of current-state systems and data environments to inform capability gap assessments, identifying modernization opportunities and technology readiness considerations for agency leadership.

Enterprise Data Architect / Delivery Lead - Accenture Federal Services

(2022-01)

Department of Veteran Affairs – Enterprise Data Architect / Delivery Lead

• Provided strategic architectural guidance for data modernization and loan processing operations.
• Built AI-enabled data lake solutions for ingestion, quality automation, and predictive analytics.
• Implemented Databricks Lakehouse as an enterprise data processing platform, integrating structured and unstructured sources with Delta Lake.
• Created Databricks Auto Loader pipelines with streaming ingestion using Structured Streaming for near-real-time data availability.
• Designed secure and governed datasets using Unity Catalog hierarchical namespaces, tags, and fine-grained access controls.
• Developed ML models for sensitive data classification using MLflow experiment tracking, hyperparameter tuning, and automated model lineage.
• Orchestrated ETL and ML pipelines using Databricks Workflows, aligning with enterprise requirements.

MBA