Restaurant Review RAG Chatbot
→
Summary
Engineered a Retrieval-Augmented Generation (RAG) chatbot that leverages the Google Maps API for user geocoding and the Yelp API for reviews, delivering real-time suggestions for the top 5 nearby restaurants.
About
Highly accomplished Data and GenAI Engineer with proven expertise in designing and optimizing large-scale data systems across finance and technology sectors. Proficient in building robust ETL and streaming pipelines using Spark, Kafka, and Airflow, I leverage LangChain and LLMs to automate business logic and deliver insightful analytics. Demonstrated success in cost reduction, latency reduction, and enabling self-service analytics, consistently driving measurable impact.
Work
Intuit
|AI Engineer
Unknown, Unknown, US
→
Summary
As an AI Engineer at Intuit, Priyank spearheaded the development of AI-powered solutions, optimizing compliance and customer feedback systems to drive significant operational efficiencies.
Highlights
Spearheaded development of an AI-powered Course Compliance Assurance Monitor, translating natural language rules into Spark SQL to auto-detect violations and generate alerts, reducing audit effort by 40%.
Engineered a Voice of Customer classification system using LangChain and LangGraph, reducing classification categories by 70% to enhance ticketing accuracy and deliver actionable insights.
Developed REST and GraphQL APIs to classify feedback, expose metrics, and store results in DynamoDB, powering internal dashboards for product teams.
Orchestrated Spark Streaming and Glue ETL pipelines to process millions of daily records, enabling self-serve analytics for Data and Business Analysts via AWS Step Functions.
Designed and built Qlik Sense dashboards for security compliance and company-wide events, integrating AWS Athena & Databricks data to improve visibility for 10,000+ employees.
Citi Bank
|Data Engineer
Unknown, Unknown, US
→
Summary
As a Data Engineer at Citi Bank, Priyank constructed and optimized data pipelines and models, enhancing data quality, reducing latency, and ensuring compliance across critical financial systems.
Highlights
Constructed Scala Spark Streaming jobs and Apache Airflow workflows, reducing data ingestion latency by 50%.
Designed and maintained DBT models to translate legacy schemas, improving data quality and data lineage across analytics pipelines.
Processed hundreds of thousands of Kafka event streams per second, applying Spark SQL transformations for compliance reporting.
Optimized Couchbase persistence, boosting query performance by 45% to enable faster insights for global risk teams.
Wells Fargo
|Data Engineer
Unknown, Unknown, US
→
Summary
As a Data Engineer at Wells Fargo, Priyank developed and engineered robust ETL pipelines and data models, migrating terabytes of data, reducing infrastructure costs, and accelerating ML model deployment.
Highlights
Developed Python and Spark ETL pipelines to migrate terabytes of on-premise data to AWS/Databricks, cutting infrastructure costs by 25%.
Engineered real-time data ingestion with Kafka, unifying multi-source data for consistent reporting across 50+ downstream systems.
Created conceptual and physical data models for ML pipelines, reducing feature processing time by 35% and accelerating model deployment.
Education
University of Texas at Arlington
Arlington, Texas, United States of America
→
M.S.
Computer Science
Amity University
Unknown, Unknown, India
→
B.Tech
Computer Science
Skills
Languages & Frameworks
Python, Scala, Java, SQL, Spark, PySpark, Airflow, DBT, REST APIs.
Data & Cloud Platforms
AWS (S3, DynamoDB, Athena, Redshift, Glue), GCP, Databricks, Kafka, Snowflake, Couchbase, PostgreSQL, Oracle.
GenAI/ML
LangChain, LangGraph, RAG, Vector DBs, LLM Apps, Prompt Engineering.
Visualization
Qlik Sense, Superset, Power BI.
Projects
Multi-Modal Image Search Engine
→
Summary
Developed a multi-modal image search engine leveraging advanced models to achieve high-speed query responses.
AI Book Classifier
→
Summary
Automated book genre classification using ReactJS + Python backend, cutting manual labeling effort by 80%.