Chief Technology Officer / India
The narrower conclusion was still a consequential decision
- An instrument-making business
- 410 employees
- ₹208 crore company revenue for FY ending March 2026
- A defined technical remit
- 68 R&D colleagues
- Five heads, ₹14.6 crore FY2026–27 expense and separate ₹6.2 crore laboratory capital
- Measurement to product judgement
- 29 years
- Five appointments from July 1997; CTO since January 2022
Decisions & consequences
01Withdrawing my release support when provenance did not establish fitness
In 2023 I supported an instrument release after reviewing its calibration documentation. I had taken a valid account of reference and process as sufficient reassurance about the proposed measurement use. Noor's quality team challenged that inference. The question was not whether the record existed, but whether the result, including its uncertainty and conditions, could support the decision for which the instrument was being offered. My approval had moved ahead of that explanation.
The commercial team had a reasonable concern: further work would delay a product already presented as close to readiness. One proposal was to retain the release timetable and add general qualifications to the supporting material. That would not resolve a specific unanswered question about the intended use. I withdrew my support for the affected release and requested a bounded review led by qualified engineering and quality owners. We retained the original reasoning so that the review was not merely an exercise in producing a different report after the fact.
The ₹64 lakh assessment considered twenty-four qualification results. Eighteen supported the stated purpose; six remained indeterminate. Indeterminate did not mean that the product had been proved defective, but it also could not be reported as evidence of suitability. The unresolved results stayed in the release account, with further work defined by the responsible specialists. I did not ask the team to select only the favourable results or imply that spending money on additional review had itself made the product ready.
The outcome was a narrower, more defensible management position, not a completed universal approval. The affected release remained held pending the relevant evidence. We retained the ₹64 lakh expense and the commercial consequence of waiting. I explained my earlier shortcut to the executive team and directors: I had allowed the authority of a document to stand in for a question about use. The lesson was to connect each proposed product claim to the measurement decision it was intended to support, and to allow independent quality challenge before my own confidence became part of the sales account.
Keeping seven hundred and twenty test hours out of a five-year promise
An endurance programme in 2024 examined thirty-six prototypes over a planned 720-hour laboratory exercise. Thirty-two completed the exercise and four stopped before completion. The development team had gathered useful observations, but a draft commercial presentation described them as support for a five-year product-life proposition. I had to separate the value of the test from the much larger conclusion being attached to its duration.
The argument for the broader language was understandable. Customers wanted a practical expectation, and the business had funded ₹1.1 crore of validation work. A simple lifetime statement seemed easier to communicate than a set of laboratory conditions. Yet the existence of accelerated or bounded testing did not provide its own justified conversion into years of field use. Qualified reliability specialists needed to examine the relevant mechanisms, assumptions and evidence. The four stopped prototypes also belonged in that examination rather than being dismissed as inconvenient exceptions.
I asked for the proposed life statement to be removed while the engineering and quality teams defined what the exercise could support. The revised account retained the sample count, completed duration, stopped samples and conditions. It also made clear which further investigation would be needed before a different product claim could be considered. I did not instruct a technical formula or replace the specialists' judgement with a management preference. My responsibility was to prevent the commercial timetable from supplying an evidential bridge that the test had not supplied.
The result did not establish five years of field life. The ₹1.1 crore remained validation expenditure, and the thirty-two completions and four stops remained observations from the exercise. We could use that evidence to inform further development without overstating what had been proved. The commercial cost was a less expansive statement at that stage. I regarded that as a necessary consequence of the available evidence, while asking the team to explain the remaining work clearly enough that the company could decide whether further investment was justified.
Retaining seven missed disturbances after the bench result improved
A 2025 evaluation used eighty defined brief disturbances to examine a proposed measurement feature. The initial instrument configuration captured sixty-one and missed nineteen. A summary focused on the successful captures and the appearance of the resulting display. The missing events were more important to the product decision than the polish of that display, because the feature was being proposed for work in which a brief disturbance could matter.
The team could have narrowed the demonstration to the events the instrument handled well. That would have produced a stronger-looking presentation without explaining the limits of the proposed use. I asked qualified measurement and product engineers to retain the full exercise, investigate the misses and distinguish capture behaviour from the way an observed event was later displayed. The quality function reviewed the evidence independently. We also kept the bench conditions explicit, so that a result from a defined exercise would not become an unsupported account of every customer's electrical environment.
The redesign and evaluation cost ₹73 lakh. In the later run, seventy-three of the eighty defined events were captured and seven were still missed. That was an improvement within the exercise, but it did not establish complete coverage. I required the remaining seven to stay in the technical and management account. The team had to explain the permitted product proposition and the further work still needed, rather than replace the original denominator with a more favourable selection.
The outcome supported continued engineering under a narrower claim. It did not establish that the instrument would detect every relevant disturbance in service, nor did it authorise us to operate a customer's network. We retained the cost and the unresolved cases. My contribution was to keep the evaluation useful for a decision: a count that improved, a boundary that remained, and a commercial statement that followed both. The episode also taught me to ask what a convincing product demonstration had excluded before treating its visual clarity as evidence of measurement completeness.
Declining a component substitution that had not earned product approval
Procurement proposed an alternative component in 2026 after identifying a supplier with an attractive delivery proposition. Its nominal rating appeared suitable for a measurement assembly already in development. The purchasing opportunity was real, but it did not establish that the resulting product would meet its declared criteria under the conditions for which it was being assessed. I asked the team to treat the proposal as a qualification question before it became an approved substitution.
One option was to accept the familiar rating and resolve any difficulties during later production checks. That approach would have moved uncertainty downstream while allowing the purchasing decision to look complete. Another option was to reject the alternative immediately because the established component was familiar. Neither was an adequate technical account. Engineering and quality owners designed the assessment of the proposed assembly, while procurement retained responsibility for supplier terms and availability. Their roles informed one another without one substituting for the other's conclusion.
The ₹86 lakh qualification programme examined forty-eight assemblies under the declared test conditions. Thirty met the stated criteria and eighteen did not. We retained both groups in the account, including the expense of investigating a candidate that had not become ready for release. A successful result for some assemblies could not be treated as approval for all of them. Nor did the finding establish that the supplier's component was unsuitable for every possible application; it concerned the product and conditions we had assessed.
I declined to approve the substitution for the proposed release on that evidence. Further engineering investigation remained a separate option for management to consider, with its scope and cost to be defined. The purchasing opportunity did not become realised savings, and the ₹86 lakh was not hidden inside a future benefit claim. My role was to preserve the difference between an interesting supplier option and a qualified product decision. The broader lesson was that a familiar number on a specification can begin a useful investigation, but it cannot finish the investigation on the company's behalf.