Veeresh Anumandla on Tal
Headline
Agentic AI Automation Engineer @ Amdocs | Agentic AI + Playwright MCP + Cursor Agent | 8+ yrs SDET | 5G Charging | Building LLM-powered self-healing test frameworks | Open to AI QA Lead roles
About
I'm an Agentic AI Automation Engineer with 8+ years turning manual test workflows into intelligent, self-healing systems. Today I sit at the intersection of Test Automation and Agentic AI at Amdocs, on AT&T's 5G Charging & Billing platform — using LLMs (Cursor Agent, OpenAI GPT-4/4o, Anthropic Claude) not as a shiny add-on, but as an everyday engineering companion for every part of the loop: understanding requirements, designing tests, generating code, debugging failures, and shipping automation for teammates on the same day they ask for it. What I've built recently: • Code-less, prompt-driven UI automation tool — Cursor Agent + Playwright + TypeScript + Model Context Protocol (MCP), with LLM-powered auto-healing locators. Teammates describe a flow in plain English; the tool records, replays, self-repairs on UI changes, and caches learned selectors so the next run is 30× faster. • RAG knowledge base of test runbooks, reference data, and environment topology — so AI-generated test steps are context-aware and environment-correct instead of hallucinated. • Reusable prompt library (few-shot, chain-of-thought, role-based personas) that enforces consistent, hard-assertion-first test output across the whole team. • Self-service Test Execution Dashboard — a web UI that lets manual testers launch, monitor, and get results from automated suites without touching an IDE. Removed the "automation engineer bottleneck" for the whole team. • Agentic flows that orchestrate multi-step scenarios end-to-end — environment discovery → provisioning → usage → downstream verification — with minimal human intervention. Numbers I'm proud of: 50%+ reduction in test design effort per release, 60+ high-impact defects caught in a single feature (earned a performance bonus), and regression cycles compressed to a fraction of what they were pre-AI. Underneath the AI, the foundation is solid: Java 8/11/17, Python, TypeScript, Selenium, Playwright, REST Assured, TestNG, Jenkins, Docker, Kubernetes, Kafka. I move comfortably between designing an AI-augmented framework and getting hands-on with a pod log or a Kafka topic dump at 2 AM. What I'm looking for: AI-Powered Test Automation Lead / Architect roles where I can scale this playbook — bring Agentic AI, LLM, and self-healing automation to a larger org, mentor engineers to work AI-first, and own quality end-to-end for AI-critical products. Hyderabad / Pune / Remote. Let's talk if you're building AI-first QA teams, hiring for Agentic AI + Playwright + MCP, or curious how LLMs actually earn their keep in a live production test suite.
Work experience
AI Automation Engineer at Amdocs Dvci Private Limited.,Pune, Nov 2022 - Present · 3 yrs 9 mos, Pune District, Client: AT&T Charging & Billing Platform (5G, Telecom) • Architected an Agentic AI test automation platform using Cursor Agent, OpenAI GPT-4/4o, and Anthropic Claude — LLMs translate design specs and API schemas directly into runnable automation, cutting per-release test design effort by 50%+. • Built a code-less, prompt-driven UI automation tool from scratch: Playwright + TypeScript + Model Context Protocol (MCP) + LLM auto-healing locators. Teammates describe UI flows in plain English; the tool records, replays, self-heals on DOM changes, and caches learned selectors — reducing UI automation authoring time from hours to minutes. • Designed a Retrieval-Augmented Generation (RAG) knowledge base of test runbooks, reference data, and environment topology — enables context-aware, environment-correct AI-generated test steps and self-service debugging. • Authored a reusable Prompt Engineering library (few-shot, chain-of-thought, role-based personas) that standardizes AI output across activation, usage, and billing test flows. • Delivered a Self-Service Test Execution Dashboard — web UI that lets manual testers launch and monitor automated test runs without an IDE. • Orchestrated agentic multi-step flows end-to-end: environment discovery → subscriber provisioning → usage consumption → downstream Kafka/record verification — with minimal human intervention. • Built REST API automation for 5G charging microservices deployed on Kubernetes; integrated AI-augmented suites into Jenkins CI/CD with Docker. • Performed defect triage and RCA using Splunk, AppDynamics, and pod-level logs; drove issues to closure in JIRA. Stack: Cursor Agent, Playwright MCP, TypeScript, Java 17, Python, OpenAI GPT-4/4o, Anthropic Claude, RAG, Selenium, REST Assured, TestNG, Jenkins, Docker, Kubernetes, Kafka, Splunk, AppDynamics.
