Maneesh Maddala — Principal SDET Engineer | From Test to Trust Skip to main content
From Test to Trust TEST TRUST
Connect

Three chapters. One pursuit:
software people can trust.

Over 15+ years, my work has evolved from hands-on testing to automation architecture and building trustworthy AI systems. Today, as a Principal SDET Engineer, I build evaluation harnesses, guardrails, and human-guided agent workflows that turn quality from a final checkpoint into evidence that systems deliver real value. Across fintech, insurance, and enterprise platforms, I uncover hidden risk, challenge assumptions early, and engineer trust into how software is designed and delivered.

Maneesh Maddala, Principal SDET Engineer at Morningstar
Maneesh MaddalaPrincipal SDET Engineer · Morningstar, Inc.
Chapter I

Exploration

Following a written script only catches the bugs already expected. The failures that matter most usually live between assumptions, so I treat testing as an active technical investigation grounded in how real people use the product.

Fresh eyes

Explore what actually happens, not only what the team hopes will happen.

User first

Test interruptions, messy data, rapid actions, and real-world constraints.

Ask early

Challenge risky assumptions before they become expensive implementation.

Chapter II

Automation

Automation should shorten feedback without creating a framework only its author understands. I favor direct tools, readable code, and the smallest architecture that the whole team can maintain.

Use tools directly

Build with Playwright or Cypress without unnecessary wrapper layers.

Test at the right level

Place UI, API, visual, component, and accessibility coverage where each gives the clearest signal.

Maintain trust

Fast, stable tests and obvious failure traces keep the suite credible.

Chapter III

Trustworthy AI

AI systems are probabilistic, exposed to injection, and capable of confident error. Trust has to be engineered into the architecture through measurable evaluation, constrained access, resilience, and human review.

Evaluate continuously

Score accuracy, safety, and structure before an agent takes action.

Constrain behavior

Validate inputs, schemas, tool calls, and outputs beyond prompt wording.

Design for resilience

Use scoped tools, observable routing, and graceful provider failover.