Chief Information Officer / India
Account for the capability that remains
- Computing colleagues
- 58
- Through three heads within a 620-person business
- Annual operating expense
- ₹19.8 crore
- Separate from ₹8.4 crore approved capital allocation
- Computing experience
- 33 years
- From July 1993 to October 2026
Decisions & consequences
01Separating thirteen cancellations from twenty-eight reassignments
Orelvi's annual renewal proposal listed 186 named specialist-software subscriptions. A preliminary usage report suggested that forty-one had no current active requirement. The finance team wanted to treat that population as a cancellation opportunity. Engineering support manager Asha Vemlori objected that some demand had moved between colleagues and that the contracts did not all allow the same treatment. I agreed that the report justified a review, but not a single action applied to every seat.
My team reconciled the forty-one subscriptions with current work, planned assignments and the fictional contracts governing them. Twenty-eight could be reassigned through the permitted vendor process to colleagues with an established requirement. Thirteen were no longer required and were cancelled at their next renewal. The company did not claim a right to transfer licences between legal entities or disregard named-user conditions. Procurement documented the permitted changes, and business owners confirmed that the revised allocation supported their work.
The thirteen cancellations reduced annual subscription expense by ₹39 lakh, at ₹3 lakh per subscription. The twenty-eight reassignments were reported separately. They improved the match between paid entitlement and need, but were not counted as a second ₹84 lakh cash saving: the company still paid for those subscriptions. I rejected a presentation that added both populations into one reduction figure. A clearer account mattered more than a larger headline, particularly when the board was comparing recurring technology commitments.
There was a practical cost to the review. Application owners spent time checking requirements and some expected reassignment dates moved while procurement confirmed the process. The company retained subscriptions where future use remained justified. The outcome was a narrower but supportable renewal decision, not a general finding that low activity equals waste. I would begin the exercise earlier in the next cycle so that renewal notice dates, technical preparation and business confirmation could be considered together instead of competing during the final weeks.
Reporting the two toolchains that still could not be rebuilt
A customer enquiry prompted Orelvi to examine whether it could recreate the computing environments used for twenty-four historic engagements. The files had been retained, and the commercial team initially expected that access would therefore be straightforward. I asked for a reconstruction trial using authorised material before the company made a wider service promise. The test concerned the environment needed to revisit work, not a fresh engineering approval of the original calculation.
Nineteen environments were rebuilt in the first exercise. Five were blocked by missing or incompatible dependencies. The team described the obstacles individually rather than marking all retained files as recoverable. Commercial colleagues were concerned that this would weaken confidence in the archive. I argued that the customer needed to know what the business could actually provide; a familiar folder structure was insufficient evidence of a usable toolchain.
I approved ₹62 lakh within the capital allocation for a bounded correction programme, subject to procurement confirming the necessary rights and business owners approving the purpose. The programme recovered three of the five blocked environments. Two remained unavailable because the required dependencies could not be restored within the authorised route. The final account was twenty-two rebuilt and two blocked, with the limitation retained in the service record. No unsupported alternative was used merely to close the count.
The result improved the company's understanding of its retained work but did not establish indefinite future availability. Nor did a reproduced environment prove that an engineering model was correct or suitable for a new use. Those questions remained with qualified owners. I changed the closeout discussion for future engagements to include the intended retention purpose, dependency record and likely cost of maintaining access. My lesson was that preservation should be designed around a business obligation and lawful use, rather than inferred from the continued existence of files.
Choosing three workstation configurations instead of one
A proposed refresh for sixty workstations used a uniform ₹3 lakh specification, producing a ₹1.80 crore purchase estimate. The proposal was easy to administer and gave every recipient the same headline capability. I questioned whether the additional components addressed the work each group performed. Several teams wanted more graphics capacity, while support evidence suggested that other delays arose from memory pressure or processor-heavy tasks with little graphics demand.
I asked application owners to define representative authorised jobs and compare supported configurations under recorded conditions. The exercise was not a competition between employees, and it did not use confidential customer material without clearance. It considered whether a task could run reliably, the resource that constrained it and the support consequences of carrying different equipment. Future demand and replacement availability remained judgement calls; a short test could not settle them completely.
The approved mix comprised twenty workstations at ₹1.5 lakh, twenty-five at ₹2.4 lakh and fifteen at ₹2.8 lakh. The total was ₹1.32 crore: ₹30 lakh, ₹60 lakh and ₹42 lakh respectively. That was ₹48 lakh below the uniform proposal. Procurement confirmed the configuration and support assumptions before the order. I kept the comparison at purchase-cost level rather than converting test-time differences into a claimed reduction in salary expense or promising a universal speed improvement.
The decision introduced some operational complexity. Support colleagues had to maintain three approved configurations and make allocation reasons understandable to staff who associated a larger machine with seniority. I supported the team in explaining the workload basis and provided a route to review a changed requirement. The outcome was a more specific investment account, not a declaration that cheaper equipment is always better. I would revisit the configurations as tasks changed and retain evidence of why the original mix had been selected.
Keeping two quiet interfaces until annual work had moved
Orelvi planned to retire an internal application supporting project administration. Its dependency list contained nineteen interfaces. Twelve had recorded activity during the preceding ninety days; seven did not. A technical proposal treated the seven quiet interfaces as candidates for immediate removal. I asked the team to identify the consumer owner for each one before accepting that absence from the activity window meant absence of a business obligation.
The owners confirmed that five interfaces were no longer needed and approved their closure. Two supported annual reconciliation work that had not occurred within the ninety-day window. Removing them would have interrupted a process whose next run was still months away. The discovery did not imply that the activity logs were wrong. It showed that the observation period answered a narrower question than the retirement proposal assumed.
I retained those two interfaces for a four-month transition, costing ₹12 lakh in additional support expense, and authorised a ₹26 lakh replacement within the capital programme. Consumer owners specified the required outputs, reconciled a trial and accepted the replacement route before the old connections were removed. The twelve active interfaces followed their separately planned migration; the case did not claim that all nineteen were resolved by the same intervention. Five early closures and two later replacements were reported as distinct decisions.
The delay frustrated the application team, which wanted to stop maintaining the older service. I acknowledged that cost rather than describing the pause as free caution. The alternative was to transfer an unrecognised interruption into another team's annual work. My next change was to require consumer confirmation and a look across the business cycle at the start of retirement planning. Logs remained useful evidence, but they no longer served as the sole authority to withdraw an interface whose purpose had not been checked.