The journey so far
Miriam Qelvamri
Chief Technology Officer · India
I grew up in Kottayam and am now based in Puducherry.
26 years
Engineering career
Stage systems through virtual-production CTO
₹91 crore
FY2026 revenue
₹57 crore stage services and ₹34 crore previsualisation
158
Enterprise employees
March 2026 headcount
44
Technology colleagues
Five leads in the CTO function
The compass
A virtual stage joins an imagined scene to a physical camera and the people working around it. I want the engineering account to say where that relationship has been tested and where a different approach is needed. A beautiful still image is not permission to promise every camera move, lens choice or set change. My own dimensional mistake taught me to make the limits of a preview visible before another team builds its plan around them.
Expeditions
01
In 2019, as engineering director, I approved an import workflow for digital set objects. The scene looked plausible, and I accepted that appearance without requiring a sufficient check of the dimensions the next team would use. A later review examined twenty-four objects: nineteen matched their intended measurements and five did not because an import-unit assumption had been applied incorrectly. The problem was not that every object was defective or that a physical set had already failed. It was that my release decision had allowed visual plausibility to stand in for dimensional evidence. The production team wanted to preserve the planned preparation sequence. Rechecking every affected relationship would consume time, and some colleagues argued that experienced scene artists could correct obvious discrepancies as they worked. That was a serious practical argument, especially where the image already communicated the intended composition. I disagreed with relying on informal correction because a downstream planner might use a dimension without knowing that it rested on an assumption. I had authorised the workflow and needed to fund its repair, rather than place the burden on whoever noticed a mismatch later. I approved ₹29 lakh for verification and rework. Engineering and creative colleagues compared the five exceptions with the intended dimensional references. Four could be corrected; one needed replacement. We retained the nineteen supported objects and documented the changes to the others, rather than claiming that the whole scene had been rebuilt. The resulting set contained twenty-three supported original or corrected objects and one replacement. The production owner then assessed the revised material for its intended use, with physical construction and safety approval remaining outside my technical sign-off. We gave up the convenience of treating a successful import as a complete release test. The consequence was additional expense and preparation work, but also a clearer boundary around what the preview could support. I cannot claim an accident prevented, a quantified downstream saving or universal correctness in later scenes. The learning was narrower and more durable: if another team will use a digital object as a measurement, the workflow must establish that measurement explicitly. I now ask which decisions depend on the scene, not only whether its first presentation looks coherent.
02
In 2022 we reviewed thirty-five planned camera moves for a virtual-stage assignment. Twenty-seven maintained the agreed relationship between the captured view and the virtual scene. Eight did not. The exceptions required examination of the camera, lens and tracking configuration, including the calibration relationship used by the scene. A still frame had not exposed the issue sufficiently. The question was whether the intended movement could be supported in that arrangement, not whether camera tracking was useful in general or whether one supplier's product was at fault. Meera, our creative lead, wanted to preserve the movement because it was part of the scene's visual intention. The stage team also needed a dependable preparation sequence rather than an open-ended technical investigation. One option was to constrain all eight moves immediately. Another was to assume that a familiar calibration file remained applicable and proceed. I proposed a bounded correction programme, with explicit acceptance of the possibility that some moves would still need another arrangement. That required the commercial team to avoid promising full restoration before engineering had completed the work. I authorised ₹58 lakh for configuration review, calibration work and rehearsed verification. Engineers checked the relationship used for the relevant camera and lens settings and tested the actual movement again. Six of the eight exceptions became supportable under the documented conditions. Two remained outside that arrangement and were returned to creative and production owners for an alternative. We therefore had thirty-three supported moves and two unresolved against the original intention, not a universal certificate for the stage. Changes to the configuration would require their own consideration rather than inheriting the result automatically. The cost included technical effort and a less flexible production promise. Meera retained the decision about whether an alternative movement still achieved the intended scene; engineering did not declare that a different shot was artistically equivalent. No real footage, customer or vendor performance is represented by this example. My learning was to connect a capability statement to the configuration and movement actually examined. A successful static view can be a useful starting point, but the executive decision must fund evidence about the behaviour on which the commissioned production depends, including the limitations that survive the test.
03
During a 2023 production-preparation exercise, unwanted visible patterns appeared in seven of eighteen tested camera-and-background setups. Eleven were acceptable for the intended picture. The issue concerned the interaction of the display and the capture arrangement, rather than a claim that seven panels were defective. The team needed to examine the recorded result for the planned shot. A general statement about display quality or an attractive preview could not settle whether that particular lens, position and framing produced an acceptable image. Creative colleagues had chosen those setups for compositional reasons. Altering the camera position or framing could change the scene, so a technical recommendation carried an artistic cost. Commercial colleagues wanted a simple answer about whether the stage was available. I resisted both a blanket rejection of the display and a blanket assurance that the issue could be removed without consequence. The useful choice was a funded examination of alternatives, with the production owner able to reject a technically clean arrangement if it no longer served the creative intention. The ₹44 lakh programme covered testing and agreed setup adjustments. Three of the seven exceptions became acceptable to the production within the changed conditions. Four did not and needed another arrangement. The result was fourteen supported setups and four excluded from the original promise. We recorded the relevant conditions rather than publishing a universal corrective setting. The case does not claim a particular pixel pitch, camera specification or technique will remove every visible pattern, nor that a simulation can replace examination of the captured image in all circumstances. We sacrificed the simplicity of selling one undifferentiated capability. The team had to explain which shots remained supportable, which creative changes had been accepted and where a different approach was still required. That was less convenient than reporting that the display had passed, but more useful to the people preparing the production. I learned to ask who bears the cost of a technical adjustment and whether that person has actually accepted it. The engineering outcome and the artistic outcome are connected, yet neither can be inferred solely from the other. My responsibility was to make that negotiation explicit before an unsupported shot entered the schedule.
04
In 2024 a rehearsal examined twelve planned camera moves after the intended set dressing was in place. Nine maintained the required tracking continuity; three encountered obstruction that prevented the arrangement from supporting the intended move. The technical setup had been assessed earlier, but the physical scene had changed. That difference mattered. I did not treat the result as proof of equipment failure, nor as a reason for engineers to remove creative elements without agreement. The production needed a choice that respected both the picture and the tracking conditions. Meera argued that the set dressing contributed meaningfully to the scene and could not be treated as disposable clutter. Engineers explained the visibility constraints of the arrangement they were supporting. The stage team wanted to avoid a series of late changes that would make rehearsal results obsolete. We considered relocating elements, revising movement and using a different production method. I asked the teams to work on two achievable revisions while preserving an explicit alternative for the third move, rather than turning an incomplete technical solution into a promise that creative colleagues would discover on the day. I authorised ₹39 lakh for joint rehearsal and configuration work. Two of the three interrupted moves became supportable after agreed changes, bringing the supported total to eleven. The remaining scene used a locked-camera alternative approved by the production owner. That alternative was a changed creative and technical choice, not evidence that the original moving shot had succeeded. We retained the distinction in the production record. Physical rigging, access and safety decisions continued to require their proper professional approvals; my technology decision did not replace them. The programme cost time and required a compromise in one scene. It also made clear why readiness should be assessed with the relevant physical arrangement, not only in an empty or earlier version of the stage. I learned to ask which changes can invalidate the assumptions behind a rehearsal and who will trigger another review. That does not mean every minor adjustment requires the same exercise. It means the team needs a reasoned account of material changes, including those introduced for creative reasons. A supported camera move belongs to a configuration, and the executive promise should preserve that dependency.
05
A 2025 review found that nineteen of twenty-four planned scenes met the agreed rendering-time requirement on the intended arrangement. Five did not. A ₹1.6 crore hardware expansion was proposed as the next step. It was plausible that additional capacity could help, but the proposal did not yet distinguish workload that could be simplified acceptably from scenes whose creative requirements justified a different method. I wanted that distinction before committing capital. The issue was the time available to produce each required image, not a claim about distribution bitrate or a network fault. The engineering team favoured a clearer capacity margin. Creative colleagues were concerned that optimisation might become a euphemism for losing visual elements they considered essential. Finance wanted to know whether the expansion solved a recurring requirement or only the current exceptions. I asked for a bounded programme that preserved creative acceptance and reported the scenes it could not bring within the requirement. The choice was not to dismiss equipment investment permanently. It was to avoid treating a purchase as the answer before understanding the production work the purchase was expected to support. I authorised ₹62 lakh for scene analysis, optimisation and retesting. Three of the five exceptions became supportable with changes accepted by the production owner. Two retained an alternative rendering method. The final account therefore showed twenty-two scenes within the agreed real-time arrangement and two outside it. We deferred the ₹1.6 crore proposal for a later capacity decision informed by future workload. The difference between those amounts was not reported as realised savings: the smaller programme bought different work, and later expansion might still be justified. The outcome required two kinds of restraint. Engineering could not remove important creative content merely to pass a timing test, and commercial colleagues could not claim all twenty-four scenes worked through the preferred method. We preserved the alternatives and their implications for preparation. My learning was to define the service capability being purchased before comparing capital options. A scene that looks good when rendered without the same timing constraint is not automatically evidence for real-time delivery. Investment judgement should make that difference understandable to directors without asking them to adjudicate individual technical settings or creative preferences.
Terrain
Languages
Fellow travellers
In this invented capital discussion, Miriam kept the two rendering alternatives visible and did not present deferred hardware as a realised saving. That made the proposal easier to question on its actual service requirement rather than on a headline comparison between unlike amounts.
The fictional setup review left creative acceptance with the production. Miriam did not assume that a cleaner technical image was automatically the right picture, and the four excluded arrangements remained part of the account rather than disappearing from the result.
Dispatches
A plausible scene can quietly acquire authority it was never checked to carry. My dimensional mistake changed how I decide which uses of a preview the engineering record must support.
Read the essayChanging a camera setup can remove a technical problem while changing the scene itself. Engineering leadership has to make that exchange visible and leave creative acceptance with its proper owner.
Read the essayA larger rendering estate can be a sensible investment. It is still a poor substitute for understanding which production requirements the current arrangement cannot meet and what alternatives preserve the commission.
Read the essayAn invitation
I would explore a first governance contribution in a creative technical-service enterprise. I have executive board-presentation experience, no statutory appointment and preparation still to complete; eighteen days is an initial annual capacity estimate rather than an unconditional availability promise.
office@miriamqelvamri.example