Asil Kamepalli

Audience, Campaign & Real-Time Media Intelligence Platform

Modernizing campaign, audience, CRM, clickstream, impression, conversion, and reference-data processing for near-real-time analytics, attribution, segmentation, optimization, and AI-enabled use cases.

• Designed an Azure-first lakehouse platform processing 2+ TB of daily batch and streaming data into governed Bronze, Silver, and Gold data products.
• Engineered near-real-time Kafka/Event Hubs pipelines with Spark Structured Streaming handling 50M+ daily events with checkpointing, watermarking, replay controls, and idempotent writes.
• Developed reusable PySpark and Delta Lake pipelines for deduplication, late-arriving data, schema evolution, SCD Type 2 history, incremental MERGE processing, and standardized business-rule execution.
• Modeled campaign, audience, attribution, and performance marts in Snowflake with dbt tests, snapshots, documentation, clustering, and workload-specific warehouse sizing.
• Optimized Spark workloads with AQE, partition pruning, broadcast joins, skew mitigation, compaction, caching, executor tuning, and cluster right-sizing, reducing runtime by 30% and compute by 25%.
• Established platform observability with Azure Monitor, structured logs, pipeline metrics, freshness/SLA dashboards, automated alerts, and runbooks for retries, backfills, late data, dependency failures, and incident triage.
• Built governed unstructured-data pipelines for RAG workloads covering extraction, normalization, chunking, metadata/version enrichment, embeddings, vector indexing, and security-filtered retrieval datasets.
• Partnered with analytics, data science, product, and business teams to define data contracts, resolve ambiguous source logic, review architecture, troubleshoot production issues, and mentor engineers on Spark, SQL, testing, and DataOps.

Enterprise Claims, Member & Provider Data Modernization

Migrated legacy payer-data workloads to a governed AWS/Snowflake platform supporting claims, eligibility, member, provider, pharmacy, finance, and operational reporting.

• Modernized healthcare integration workloads into an AWS-based platform using S3, Glue, PySpark, Airflow, Snowflake, and dbt across raw, standardized, and consumption-ready layers.
• Built reusable ingestion frameworks for database extracts, REST APIs, partner files, and CDC feeds with encryption, schema capture, audit columns, retries, quarantine handling, and recoverable incremental loads.
• Developed PySpark and AWS Glue transformations for high-volume claims, eligibility, member, provider, and pharmacy datasets of 50M+ records while handling duplicates, source corrections, and late-arriving records.
• Designed Snowflake dimensional models and dbt layers with fact/dimension structures, conformed dimensions, SCD Type 2 processing, incremental strategies, snapshots, automated tests, and documentation.
• Implemented source-to-target reconciliation for row counts, claim/payment totals, referential integrity, duplicate detection, freshness, and source-target variances before analytics certification.
• Improved Spark and Snowflake performance across multi-terabyte tables, achieving 35% latency reduction and 30% compute cost savings.
• Owned production support for failed loads and unexplained data discrepancies, tracing issues across source extracts, orchestration, transformations, and warehouse models, then converting root causes into preventive controls.
• Applied least-privilege IAM, KMS encryption, secrets management, audit logging, and traceable lineage to sensitive healthcare datasets.

Global Sales, Inventory & Supply Chain Analytics Platform

Integrated ERP, product, inventory, sales, distribution, and operational data to support trusted KPI reporting, fulfillment visibility, and analytics.

• Built Python and SQL ETL pipelines integrating ERP, product, inventory, sales-order, distribution, REST API, CSV, and JSON sources, processing 8M+ records daily.
• Created reusable ingestion modules for connectivity, file handling, incremental watermarks, audit logging, exception handling, and restartability.
• Developed SQL and PySpark transformations using CTEs, window functions, aggregations, joins, deduplication, standardization, derived attributes, and business-rule validation.
• Designed star-schema marts with fact and dimension tables, surrogate keys, date dimensions, and historical attribute management for consistent sales, inventory, and operations KPIs.
• Implemented source-to-target reconciliation, null/duplicate checks, schema validation, exception reporting, and controlled reprocessing to prevent bad data from reaching scheduled reports.
• Optimized Parquet layouts and relational queries through partitioning, file sizing, indexing, predicate filtering, query rewrites, and staging-table design, reducing query latency by 45%.
• Automated recurring batch jobs with dependency checks, retries, file-arrival validation, notifications, and restart procedures, improving reliability over manual scripts.