Chief Operating Officer — Machine-Learning Infrastructure Stack

The mandate

Operating complexity has outgrown the governance used to build a machine-learning infrastructure stack. Teams responsible for capacity, data, platforms, delivery and controls meet their local objectives, but customer journeys still cross too many hand-offs. Responsible-AI requirements add consequential evidence and decision points to those journeys. The board has concluded that incremental coordination will not solve the problem and is creating a Chief Operating Officer role with end-to-end authority.

The business perimeter includes approximately ₹1,200 crore in AI product and services revenue and about 1,000 employees and material partners across India, Bengaluru and the wider operating region. It spans platform operations, infrastructure, programme delivery, capacity, service management, supplier relationships and operating controls. Demand varies by product and customer, yet scarce resources are often allocated through historic ownership rather than whole-system priorities.

The COO will build an operating model in which delivery reliability, productivity and responsible-AI evidence reinforce each other. Faster throughput achieved by moving risk downstream is not improvement. Nor is a control system successful if it depends on manual reconciliation and specialist intervention at every release. The goal is a small number of visible flows with accountable owners, usable standards and trustworthy measures.

Why this seat is open

This is a newly created urgent appointment, established because distributed ownership cannot carry the control build through the next phase. Interim forums safeguard current decisions but are not a substitute for one permanent executive. The board plans to progress from qualified shortlist to offer within six to eight weeks. The hybrid Bengaluru role is accountable to the Group Chief Executive or a designated executive committee sponsor.

What you will own

You will map the principal flows from customer commitment through data, infrastructure, model release, evaluation and live service. Each flow needs one executive owner, explicit service expectations and measurable constraints. Functional leaders retain expertise and people authority, but disputed priorities and hand-offs must be resolved through the end-to-end outcome rather than hierarchy or negotiation stamina.

Capacity and productivity require a common language. You will reconcile demand, compute, specialist skills, third-party supply and delivery commitments, differentiating genuine constraint from local buffering. Work-in-progress limits, planning horizons and escalation triggers should support decisions. Cost reduction that increases incident, delay or evidence risk elsewhere in the stack will not qualify as productivity.

The responsible-AI control build must enter everyday operations. Together with technology, product and risk leaders, you will define where evidence is created, who approves exceptions and how monitoring responds to change. Controls should be automated where sensible and tested through real releases. Customer and regulatory obligations need traceability from contract to operation.

You will also lead organisation and cadence changes across 1,000 employees and partners. Assess the leadership team, appoint missing flow owners and develop successors. Consolidate reviews that repeat the same data with different narratives. Suppliers should be integrated into service, capacity and resilience governance, with commercial consequences for unmanaged dependency.

The first 12 months

During the first 90 days, establish an end-to-end baseline using representative customer and platform flows. Quantify queue time, rework, incidents, capacity constraints and control evidence. Meet critical customers and partners, assess direct reports and stabilise urgent delivery or responsible-AI issues. Agree a board scorecard and the few decisions needed to unlock the initial operating redesign.

Between months four and nine, introduce flow ownership, align planning and remove the most costly hand-offs. Implement responsible-AI gates in priority release paths, fill leadership gaps and renegotiate material capacity or supplier issues. Demonstrate a measurable gain in reliability or throughput on at least one important flow without weakening assurance.

By year end, delivery performance and productivity should be improving across several flows, supported by clear end-to-end accountability. The following year’s resource plan must reconcile customer commitments, infrastructure economics, control capacity and people. Present a three-year operating case with explicit fallback actions if demand, talent or platform performance departs from assumptions.

What the board will measure

Annual delivery should stay within 10% of the approved case, with forward visibility of variance. Three consecutive quarterly forecasts must join revenue, cash, delivery, infrastructure capacity, customer demand and workforce assumptions. A selected cause of operating complexity should improve quantitatively from a verified baseline under an accountable data owner.

The most consequential delivery and control weaknesses need closure on agreed dates and proof that fixes survive subsequent releases. Critical-talent retention should meet or exceed 90%, and 70% of COO direct reports should have ready-now cover. No significant surprise may bypass governance, and severe escalations cannot remain undecided beyond 30 days.

The person

You are a COO, EVP Operations or Business Operations President with 18–22 years in AI, enterprise software, cloud, data infrastructure, analytics, applied research or an adjacent operation. You have owned at least ₹1,250 crore in P&L, budget, book or accountable portfolio and led at least 1,000 people.

Your evidence includes simplifying a technically complex operating model while introducing material controls. You can quantify movement in reliability, productivity, cost and customer outcomes and explain the trade-offs behind it. The board will test a moment when functional optimisation harmed the system and how you restored end-to-end ownership. References must distinguish your choices from broader transformation activity.

Compensation and terms

The anticipated package is ₹3.2–4.6 crore fixed plus performance variable and LTI, calibrated to final scope and current mix. Long-term participation follows standard vesting and performance conditions. Notice up to six months can be managed. The COO will have regular exposure to the group board and relevant risk and people committees.

Confidentiality

Only candidates and the organisation that confirm mutual relevance will exchange identifying details under formal confidentiality. Market, scale and operating conditions are composite and should not be interpreted as a coded account of a particular enterprise.