Pratyaksh Chanda

AI Backend Engineer with 2+ years at BNY (Bank of New York) building production-grade distributed systems and agentic AI automation for enterprise engineering teams. Delivered measurable impact end-to-end — a 50% cut in incident response latency via a Kafka-based alert engine, a 35% reduction in tool-switching overhead through fine-tuned LLMs, and a fully agentic multi-agent platform for autonomous IT workflow execution. Deep expertise in event-driven architecture (Kafka), distributed caching (Hazelcast), microservices (Spring Boot), RAG pipelines, and LLM fine-tuning.

AI Backend Engineer at BNY (Bank of New York) (2024-07 – Present)

Enterprise Digital Assistant Platform — Agentic AI System

• Built a Kafka-based real-time alert engine for asynchronous event streaming, cutting incident response latency by 50% and peak-load latency by 30% via multi-threading and connection pooling.
• Fine-tuned LLMs to generate contextual, actionable responses for engineering teams, cutting tool-switching overhead by 35% and speeding up issue triage.
• Designed a fully agentic, multi-agent architecture with specialized "Digital Engineer" agents coordinated by a central controller for autonomous planning and parallel task execution across IT workflows.
• Built an Auto-MR (Merge Request) Reviewer agent using static analysis and LLM-based reasoning over code diffs, catching ~28% of risky merge requests pre-review and cutting review time by 20%.
• Architected Microsoft Teams Digital Assistant bots and integrated Jira, GitLab, ServiceNow, and Confluence via GraphQL/REST APIs for unified, context-aware operational insights.
• Configured a Hazelcast distributed cache with rolling-based context management for persistent conversation history, and built an MCP server integrating WindSurf AI with internal tooling for seamless IDE–AI collaboration.
• Built autonomous AI tools capable of independently identifying, triaging, and resolving recurring operational issues, reducing manual intervention by 25%.
• Built an AI automation flow for vulnerability remediation — ingesting VMAD portal findings, auto-triaging by severity, and notifying developers in real time via MS Teams, Slack, and Outlook — reducing mean-time-to-remediation by 35%.
• Developed an AI-driven pipeline remediation agent that auto-resolves recurring CI/CD failures (build errors, SonarQube gate failures) via rootcause analysis, with real-time alerts via MS Teams, Slack, and Outlook — cutting MTTR by 30%.

AI Backend Engineer at BNY (Bank of New York) (2024-07 – Present)

NexGen Test Automation Framework

• Developed a Python-based test automation framework on Robot Framework, adopted by 8+ teams across BrowserStack, local, and App Engine environments.
• Built reusable keyword libraries with AI-assisted keyword discovery and automated email notifications for test results, speeding up script authoring and cutting time-to-resolution.

Software Engineering Intern at BNY (Bank of New York) (2023-05 – 2023-07)

• Engineered multi-stage GitLab CI/CD pipelines (Build → SonarQube → Dev → QA → Prod) using BNY's shared templates for enterprise-wide standardization.
• Integrated FastPath AutoRFC, enabling single-day production deployments with only one approval and significantly shortening release cycles.

B.Tech in Electrical Engineering – Indian Institute of Technology Roorkee (2020-11 – 2024-05)