Avinash Tiwari

Architected and delivered a production enterprise AI agent platform using LLMs and Model Context Protocol (MCP) tool integrations, securely connecting agents to Jira, Confluence, BigQuery, Cloud Storage, and internal REST APIs. Built a Workspace Assistant for natural-language issue tracking, documentation retrieval, and analytical queries, and a Buyer Engagement agent using Google Agent Development Kit (ADK) to personalize dealer communications and automate sales engagement workflows. Designed and built an end-to-end enterprise RAG system including document ingestion, embedding generation, vector search, semantic retrieval, and prompt optimization over Confluence knowledge spaces. Implemented a 4-layer production AI guardrail framework (prompt validation, structured JSON schema enforcement, response constraints, and hallucination mitigation) aligned to internal compliance and model risk review. Also engineered cloud-native data platforms using Apache Beam, Kafka, Dataflow, and Airflow

Built a reusable Python data testing framework to standardize validation across data pipelines and shift defect detection earlier in the release cycle. Developed reporting modules in React and Node.js for an internal low-code testing platform and built a web UI to capture API requests and surface performance metrics. Improved database performance by rewriting high-cost queries and resolving related defects.

Modernized Spring Boot applications by delivering new features and resolving production defects. Deployed cloud-native applications using Cloud Foundry and implemented JUnit unit tests to validate behavior before release.