Building Agentic AI & Multi‑Agent Systems for Software Testing
Autonomous quality agents that read a JIRA story, design the scenarios, write and run the tests, analyze failures, and self-heal - keeping humans in control at the gate.
AI shift-left
improvement
automated
API and contract testing
REST endpoints checked against their OpenAPI specs, with assertions that repair themselves when fields or response shapes change.
Web and cross-browser
Playwright and Selenium suites run the same journeys across browsers, and locators heal on DOM drift instead of failing the build.
Performance and security
Load profiles and OWASP ZAP scans execute in the same pipeline as the functional checks, so regressions surface in one report.
Shipped quality in
AI Platforms
Autonomous AI Platforms
Production-grade multi-agent platforms architected for software quality engineering, automated video creation, WhatsApp conversational commerce, and global market intelligence.
Omni Quality Platform
QE Agentic Framework & Multi-Test Automation
Autonomous QE Agentic Framework that generates, executes, and self-heals tests across 6 test types (Functional Playwright, E2E, Cross-Browser Selenium, REST API, Performance, and Security OWASP ZAP) directly from JIRA acceptance criteria or live application URLs with root-cause analysis and quality gates.
StudioAgent
The Autopilot Video Studio & B-Roll Content Engine
Automates high-converting short-form video content for Instagram Reels, YouTube Shorts, and LinkedIn with zero hallucinations. Powered by an autonomous multi-agent pipeline handling research, viral scripting, compliance evaluation, rendering, and publishing.
SaarthiOne
Turn WhatsApp Chats into Bookings, Payments & Repeat Customers
An AI teammate that sells, books appointments, and collects payments 24x7 directly inside WhatsApp chats. Handles natural language inquiries, converts chats into real-time bookings, issues instant payment links, and manages customer retention.
Atlas
Global Markets & Portfolio Intelligence Engine
Monitors portfolios across India and 9 global equity markets in one clear view - providing plain-English market structures, 100+ stress-tested quantitative strategies, intraday signals, and compliance-first AI portfolio reviews.
Orbis Quant AI
Backtesting & Algo Trading Platform
Autonomous multi-agent quantitative platform designed for backtesting strategies, real-time algorithmic trading execution, portfolio optimization, and risk management across global markets.
How it works
Inside the agent loop
Four stages take a requirement from a JIRA ticket to a governed release decision - with an engineer holding the gate at the end.
01
Scenarios written for you
Point it at a JIRA story or a live URL and the platform generates hundreds of functional and adversarial scenarios - edge cases, multi-turn flows and policy violations you would never think to write by hand.
02
Six suites, one pipeline
Playwright, Selenium, REST, performance and OWASP ZAP run in parallel. When the UI drifts, locators repair themselves instead of failing the build.
03
AI judgment, deterministic proof
A model scores accuracy, groundedness and tone. Hard rules then check the answer against the ledger and the policy - so a confident-sounding response can still fail the gate.
04
Continuous regression gates
Every candidate build is replayed against the baseline corpus. Quality drift, tone shift and policy regressions surface as a diff before release, not after.
About
Engineering Lead with 15+ years of experience in Quality Engineering, Agentic AI, LLM reliability, Agile software delivery, and CI/CD.
Delivered 30-40% reduction in STLC cycle time by implementing AI-powered Shift-Left Quality Engineering and automation across Payments, Lending, Financial Services & Banking - enabling faster releases and improved software quality.
Improved platform reliability through RCA and quality governance, contributing to 99.9%+ production availability and reduced defect leakage.
Journey
Experience
A 16-year arc - from building banking interfaces by hand to pioneering AI-native quality engineering at global scale.
- Architected and launched AI platform: Omni Quality Platform (Agentic Quality Platform).
- Architected an AI-driven Agentic STLC integrating multiple LLM-based agents, DevOps and CI/CD - 30-40% reduction in testing effort, 25% faster release validation, and AI-generated test coverage.
- Designed multi-agent AI frameworks with prompt engineering, validation orchestration, and deterministic guardrails - up to 30-40% improved defect detection and zero hallucinations.
- StudioAgent - Architected & built an Autopilot Video Studio & B-Roll Engine automating short-form video generation, captions, rendering & publishing for Instagram Reels, YouTube Shorts & LinkedIn.
- SaarthiOne - Developed a WhatsApp Conversational Commerce & Bookings AI teammate operating 24x7 inside WhatsApp for automated appointment scheduling, product selling, and instant payment links.
- Atlas - Engineered a Global Market Intelligence & Portfolio Analytics Engine monitoring 10 global equity markets with 100+ quantitative strategies and AI copilot portfolio audits.
- Orbis Quant AI - Developed a Backtesting & Algo Trading Platform featuring strategy parameter sweeps, algorithmic signal execution, and portfolio risk management.
- Led Quality Engineering, Test Automation, Performance and Customer Experience across financial products spanning Lending, Payments, Insurance and Collections.
- Architected automation for APIs and Selenium UI, increasing regression automation coverage to 80%+ while significantly reducing manual effort.
- Built and led a distributed team of 30 Quality Engineers across India, Indonesia and Vietnam, establishing engineering standards and delivery governance.
- Reduced Full Cycle Testing from 54 hours to 9 hours (83%) through parallel execution, optimized regression suites and automated quality gates.
- Built CI/CD pipelines with WireMock and Grafana observability, cutting deployment validation effort by 40%+.
Hands-on engineering across quality and automation, re-grounding in delivery before stepping into AI-native quality leadership.
Co-founded an interior design execution and marketplace firm - entrepreneurship, business setup, strategy, operations and technology.
Automated an FX trade platform spanning business rules, monitoring and settlement; led end-to-end delivery and built a framework for database & message validation using JMS, shell scripting and Java.
- Worked across Tosca, Selenium, UFT/QTP, SoapUI, CA LISA (service virtualization) and RIT (GreenHat).
- Credit Suisse: automated a 15-component cash control model converting cash flows into payments.
- JPMC: automated the Mortgage Express System covering end-to-end loan processing from origination to closure.
End-to-end development and execution of SMB Project Lite using Java, JavaScript and MySQL - projects, resource allocation, role permissions, timesheets and reporting.
Developed interfaces for the FLEXCUBE retail banking core, enabling online banking across Internet, Mobile and IVR channels - fund transfers, standing instructions, term deposits, direct debit and account management.
Technical Expertise
What I bring to the table
Deep technical skills and strategic leadership to drive quality in AI-era software engineering.
End-to-end LLM evaluation using DeepEval and Guardrails - including hallucination detection, prompt robustness, semantic scoring, and AI system validation pipelines.
Architecting enterprise-grade automation across API and UI using Java, Selenium, TestNG, RestAssured and Playwright with CI/CD integration.
Designing multi-agent workflows on LangGraph with stateful DAGs, MCP servers, steering files, Guardrails, and self-correcting agent loops for autonomous testing.
Leading SDET teams across BFSI domains, defining quality governance, risk-based testing strategies and cross-functional leadership at global scale.
Improving production reliability using Datadog, Grafana, centralized logging and monitoring - ensuring quality extends well beyond deployment.
Built Orbis Quant Agents - an open-source multi-agent AI framework for quantitative finance analysis shared with the community.
Technical Deep Dives
Architecture Case Studies
In-depth breakdowns of system requirements, problem statements, execution workflows, governance guardrails, and technology stacks across our 5 flagship AI platforms.
An autonomous multi-agent quality platform that moves a requirement from intake to governed execution, failure analysis and auditable results. Features fail-closed guardrails and bounded self-healing.
A production platform connecting customer conversations to qualification, booking, payment, CRM operations and human handoff with structured intent handling and signed Razorpay payloads.
A shared platform for self-service portfolio analysis, continuous advisor monitoring and institutional market research using FACTS grounding blocks and schema-enforced outputs.
An end-to-end AI pipeline that researches, writes, evaluates, guards, renders and distributes content across social platforms using a 7-stage LangGraph state machine with SSE streaming.
A full-stack quant trading system combining strategy building, paper trading, options intelligence, broker integrations, isolated OS strategy processes, and risk gates.
Work
Company & Freelancing Projects
Key delivery work across enterprise systems, freelancing client solutions, and FinTech infrastructure.
Automated short-form video generation engine for Instagram Reels, YouTube Shorts & LinkedIn. Multi-agent pipeline handling research, viral scripting, zero-hallucination compliance guardrails, rendering, and automated publishing.
Conversational AI teammate operating 24x7 inside WhatsApp. Handles natural language inquiries, automated booking & scheduling, instant in-chat payment link generation, and customer selling.
Global markets and portfolio intelligence engine monitoring 10 equity markets. Includes 100+ quantitative backtested strategies, real-time market structure insights, and Atlas Copilot AI portfolio audits.
Autonomous multi-agent platform for quantitative strategy backtesting, algorithmic trading signal execution, portfolio optimization, and real-time risk management.
End-to-end API, web, E2E and load testing across cashloan, paylater, cicil and the loan platform. Built a CI/CD pipeline on GitLab with WireMock for third-party scenarios, a Gatling + Grafana + InfluxDB load testing pipeline, and a QA dashboard tracking automation coverage, defect rate and team status.
Automated 150 regression scripts per release cycle across 15 interlocking components that convert cash flows into payments. Built a VBA macro utility for XML test data generation and database comparison - reducing manual effort from 1 hour to 15 minutes per cycle.
Optimised and automated 4,000 test cases across MAX, MPX, MyCm, LARS and OnBase. Full end-to-end coverage of loan processing from origination to closure, with an integration framework built for end users.
Baselined regression suite run each release cycle. Automated database and batch processes including incentive calculations. Built a framework for database and report validations with end-to-end POC, estimation and execution.
Automated an FX trade platform covering business rules, monitoring and settlement. Defined requirements and led end-to-end delivery including ops portals, NFR gathering and a database & message validation framework using JMS, shell scripting and Java.
Developed interfaces for the FLEXCUBE retail banking core enabling online banking across Internet, Mobile and IVR channels. Also contributed to Citibank's JFP banking solution covering core banking modules.
Open Source
Projects on GitHub
Multi-agent AI frameworks built in the open - quantitative finance analysis with LangGraph.
A LangGraph multi-agent orchestration framework for quantitative finance - specialised analyst agents with an adversarial bull/bear debate layer, multi-perspective risk audit, and confidence-weighted portfolio synthesis.
Credentials
Certifications
Continuous learning in AI, cloud and engineering - validated by recognised programmes.
Looking for my next challenge
Open to Principal SDET, AI Transformation Lead, and LLM Testing roles - on-site, hybrid, or remote. Based in India, open to global opportunities.
Connect
Let's work together
Whether you're building AI quality teams, exploring LLM testing strategies, or looking for a technology partner - I'm always open to meaningful conversations.
Based in India. Available globally - on-site, hybrid, or remote.