Vanshaj Jain

HDFC AMC - Architected and deployed a production-grade Agentic AI orchestration system using LangGraph,
Amazon Bedrock, and Redshift as a structured knowledge base, incorporating guardrails, evaluation pipelines, and
stateful workflows for reliable enterprise AI applications.
• Designed secure, metadata-driven Airflow and DBT pipelines on Amazon Redshift for enterprise CRM and
Customer 360 data processing. Optimized workflows and parallelization reducing runtime from 8 hours to 30 minutes.
• Designed a Multi-Agent Equity Research AI system decomposing financial queries into deterministic retrieval,
computation, and LLM reasoning stages for explainable, low-hallucination investment insights, with reusable tool
abstractions (data retrieval, ratio computation, news analysis) enabling rapid platform extension.
• Tata Power Ltd - Architected a metadata-driven AWS data lake for Tata Power, unifying Salesforce, SAP, and
Wrench data through scalable AWS Glue workflows and dynamic ingestion ETL pipelines.
• Engineered incremental processing with Apache Hudi, automated orchestration, and process-level logging enhancing
scalability, auditability, and pipeline reliability.
• Tekcor4 - Modernized a 20-year-old legacy flat-file processing system into a scalable Python/Go-based platform
integrated with OracleDB, enabling dynamic multithreaded ingestion through the Chronos framework.
• Reduced processing time from 10+ hours to 1 hour and consolidated 48 legacy pipelines into 6 scalable workflows by
implementing centralized monitoring and VPC-isolated infrastructure.

• Optimized costs and machine health by identifying KPIs and developing real-time SPC dashboards.
• Reduced rejection-related costs by Rs 0.78 lacs/month through automated data visualization and RCA.

• Implemented ML methodologies like data pre-processing, training, validation to build Models for Sales prediction data.