Pulkit

Built and maintained scalable ETL/ELT pipelines using Azure Data Factory and Databricks, processing 10M+ records daily. Developed Medallion Architecture pipelines using Delta Lake and PySpark for enterprise analytics use cases. Engineered ingestion pipelines for SAP, Salesforce (SFDC), Snowflake, and flat file sources into ADLS Gen2. Implemented incremental loading and partitioning strategies reducing pipeline runtime and compute costs. Developed transformation pipelines using PySpark and SQL for curated analytics datasets. Handled late-arriving data using Delta Lake MERGE operations and lookback logic. Built Power BI dashboards and reporting datasets for business users. Built AI-powered documentation agents using Azure AI Foundry for metadata enrichment and lineage generation.

• Built and operated 15+ real-time ETL/CDC pipelines spanning ingestion to consumption using Python and FastAPI services, orchestrated through Apache Airflow, saving 10+ hours daily for finance and sales teams. • Engineered a watermark-based CDC pipeline syncing high-volume MySQL source data into a PostgreSQL shadow store, plus MySQL→PostgreSQL replication via AWS DMS with LOB support, batch-apply tuning, and CDC start-position recovery. • Deployed and operated self-hosted PowerSync sync services for real-time PostgreSQL→on-device SQLite replication across mobile clients; tuned sync rules, managed MongoDB-backed bucket storage, split replication/API containers behind Nginx, and resolved production outages to keep offline-first apps consistent at scale. • Engineered streaming CDC pipelines with Redpanda Connect to sync CRM (MySQL) data into PostgreSQL, with watchdog auto-restart services and Prometheus/Grafana monitoring for high availability. • Led data migration efforts, transferring 100+ GB to PostgreSQL with zero data loss and minimal downtime. • Architected Dial-O-Matic, a lead prioritization and ranking platform serving concurrent devices via a Redis ranking engine, denormalized PostgreSQL serving views, and trigger-based CDC. • Developed an automated self-tutoring workflow by extracting call transcriptions via ETL pipelines and analyzing them using OpenAI, delivering insights on wins, misses, promises, and improvement actions — boosting sales productivity and coaching accuracy across the team. • Built ETL pipelines for Neo4j Graph DB to process complex relationship data, and developed Python/OpenAI extraction algorithms mining 5,000+ institutions for the Find My College product. • Authored technical design documentation and deployed Vlens, an AWS-hosted analytics platform, cutting ad-hoc reporting time by 50%

• Conducted SLA, cohort, and inventory analysis using Excel and SQL, improving delivery SLA from 40% to 60% and dispatch SLA from 65% to 88%. • Designed and implemented CM1, CM2, and EBITDA dashboards in Zoho Analytics; optimized logistics to reduce last-mile delivery costs by 50%. • Built tools such as an HSN Code Converter relying on precise data tagging and taxonomy application, improving workflow efficiency. • Optimized complex SQL queries, reducing dashboard load times by 40% and accelerating data availability for business users.