Chief Technology Officer / India
A finished file still needed a narrower approval
- Technology organisation
- 41 engineers
- Four heads across publication-engineering disciplines
- Annual operating expense
- ₹8.9 crore
- FY2026–27; ₹3.6 crore capital programme separately approved
- Publication-technology career
- 30 years
- Composition support through digital-typesetting CTO responsibility
Decisions & consequences
01Revising the visual approval after sidebar material interrupted the reading sequence
In 2018 I approved converted publication samples because they looked faithful to the source. The spacing and hierarchy made the result easy to accept in a visual review. Farida examined the underlying reading sequence and found that six of eighteen samples introduced sidebar material at the wrong point in the main argument. Twelve preserved the intended order. The defect concerned how the reader encountered the material, not whether the words had vanished from the file.
The delivery sponsor wanted to preserve the visual approval and treat sequence review as a later refinement. Farida argued that the reading route was part of the commission, not an optional improvement after delivery. I accepted that challenge and reopened the six samples. Qualified editorial colleagues clarified which material belonged to the main argument and how the supplementary content should be encountered. Engineering then had a specific relationship to preserve instead of an appearance to imitate.
We used ₹52 lakh over ten weeks to correct the six and revise the acceptance process. The selected checks did not establish universal accessibility, complete conformance with every requirement or identical support across all reading systems. I wanted the corrected account to be precise about the conditions examined. The twelve already correct samples remained distinct from those requiring work, and the correction was not presented as a measured increase in reader satisfaction or commercial performance.
The trade-off was reopening a delivery account I had approved and funding work the visual review had failed to identify. The bounded result was corrected sequence in the six samples and a more specific approval method. I had to acknowledge that a persuasive page had made a narrower check appear sufficient for a broader reading promise. The lesson was to preserve the editorial relationship beneath the appearance before technical confidence becomes a commercial statement about how the publication can be used.
Restoring the return path from a note to the intended passage
In 2023 a conversion review examined twenty-eight note references. Twenty-two returned the reader to the intended passage after the note. Six returned to the wrong context. The links were active, so a test limited to whether navigation occurred could have recorded success while missing the relationship that made the note useful to the reader.
The production sponsor wanted a simpler acceptance rule based on working links. Farida asked the team to follow the full route from passage to note and back again. I supported that review because the problem was not merely a broken technical endpoint. It was a reading relationship that had changed during conversion. Qualified editors confirmed the intended context, while engineering retained responsibility for implementing the route faithfully. The author’s wording did not need alteration to compensate for our navigation error.
We used ₹44 lakh over nine weeks to correct the six references and revise the navigation checks. The twenty-two correct references remained separate from the six requiring correction. The result did not establish that every link in every publication had been examined, or that the change improved engagement or comprehension by a measured amount. It established that the selected return paths now led to the intended context under the reviewed conditions.
The trade-off was a more detailed acceptance process than checking for active links alone. I wanted the enterprise to explain that cost as part of preserving the reader’s route, not as unnecessary perfectionism after a file had passed a basic technical check. A reader following a note should not have to reconstruct where the publication meant to return them. The case reinforced a distinction that matters to technology leadership: an action can execute successfully while still carrying the wrong meaning for the person who relies on it. The engineering result must therefore be tested against the relationship it was meant to preserve.
Holding an expression rather than inventing the meaning needed to finish conversion
In 2025 we reviewed thirty-six mathematical expressions in a commissioned digital publication. Twenty-nine representations preserved the intended meaning. Seven had problems involving superscript presentation or fraction grouping. A visually plausible arrangement could still suggest the wrong relationship, so I asked that the review remain connected to qualified subject judgement rather than rely on appearance alone.
The schedule sponsor wanted all expressions marked complete so the package could proceed as one delivery. Farida supported separating technical work that could be resolved from the expression whose meaning still needed confirmation. I agreed to correct six and hold one for qualified subject review. That left thirty-five supported expressions in the selected account and one explicitly unresolved. The technology team was not authorised to invent the interpretation merely to complete the list.
We committed ₹61 lakh over twelve weeks to the correction and review process. The work examined representation in the reading conditions specified for the commission. It did not prove universal device or reading-system compatibility, and it did not create a professional certification. Client and subject owners retained acceptance of the intended content. The held item was visible in the delivery account rather than concealed within a broad statement that the publication had been converted successfully.
The trade-off was carrying an incomplete item and its scheduling consequence after most of the technical work was done. I wanted management to understand why that was a more responsible result than a complete-looking package with an unsupported interpretation. The bounded outcome was six corrected expressions and one question returned to the person qualified to answer it. Engineering leadership includes making that limit practical: funding the specialist review, explaining the consequence to the client and ensuring the system can represent a hold without treating it as an inconvenience somebody should clear by guessing.