Most validation problems I've seen are traceability problems.

The testing may exist. The issue is that the chain is not always clean enough when someone asks you to explain it end to end. These often live in different places, and that is where gaps show up:

• Requirements
• Risk
• Controls
• Evidence

A lot of my work has been about closing those gaps — making sure the system can be traced without having to rebuild the story during an audit.

"Reduce validation ambiguity before it becomes audit risk."

— Vijay Kumar, Co-founder & Validation Lead

At Complere, I work across validation, engineering, and DevOps. I prepare both CSA and CSV validation suites, and the client decides which approach they want to implement.

For each module, the process runs end to end, covering:

• User requirements (URS) and risk assessment (RA)
• Configuration specification (CS)
• Test planning (TP) and traceability matrix (TM)
• IQ/OQ test scripts and performance qualification (PQ)
• Test summary report (TSR) and validation summary report (VSR)

The same structured flow covers every module across the platform. That consistency is what makes the package defensible when someone looks at it closely.

That matters because systems do not fail in neat silos. Validation, delivery, and environments all affect each other.

AI helps with consistency and repetitive work, but it does not replace traceability or reasoning. Those are still the parts that matter most in an audit.

I've also worked on system behaviour as a controlled requirement. Modules were tested under load and performance scenarios so they remained usable in real conditions, not just technically correct in a test environment. A compliant system that is hard to use still creates problems.

At Scilife, I worked across product modules including CAPA, audits, suppliers, change control, and training. During that journey, the platform grew from 1 client to 200+ clients and from 10 users to 20,000+.

Over the years, I've worked closely with teams moving between CSV-heavy and CSA-aligned models. For me, the biggest improvement is not just efficiency — it is clarity. And clarity makes a big difference when systems are reviewed closely.