Gurugram, India · varun.py10@gmail.com · +91-8799707841 · github.com/VarunDTU · linkedin.com/in/varun-negi-93b827223
About
I build production AI assistants, analytics automation, data pipelines, and backend services with Python, Java, and AWS.
Experience
Stanza Living
Gurugram, India
Software Development Engineer I
Dec 2025 — Present
Built Support 3.0, a LangGraph assistant using pgvector-based hybrid RAG to answer resident queries from internal policies and FAQs before ticket creation.
Enabled domain agents to use conversation history, interpret images, perform controlled ticket actions, and trace response issues with LangSmith.
Migrated 200 million monthly events to an Airbyte and dbt pipeline on Amazon Redshift, reducing data-integration costs.
Built a Python framework for dbt model generation and validation, schema changes, type conversion, nested JSON transformation, and incremental processing.
Prevented overlapping transformations with FastAPI and Redis locks; triggered dbt after successful ingestion and sent Slack delay and recovery alerts.
Maintained Java and Spring Boot services across 5 business domains, serving 3,000+ daily users with zero critical downtime.
Software Development Engineer Intern
Jun — Dec 2025
Designed and deployed DataSense, a natural-language analytics agent querying 400+ Redshift tables and saving 20+ hours per week for 10+ teams.
Improved complex SQL generation through schema-aware retrieval and contextual prompts.
Improved website synchronization performance and data accuracy by 30% through query and data-processing optimization.
Tailor Talk
Mar — Jun 2025
Software Development Engineer Intern · Remote
Built Python and FastAPI agent workflows handling 3,000+ daily calendar conversations and contributing to 300+ monthly bookings.
Implemented rate limiting, reliability controls, and reporting automation, maintaining 99%+ uptime and reducing manual data entry.
Project
I.
DataSense
A Python-based data analyst agent that turns natural-language questions into SQL. Uses schema-aware retrieval to query 400+ AWS Redshift tables, saving 20+ hours per week for 10+ teams at Stanza Living.