Interim Chief Technology Officer — Generative AI Enterprise Launch

The mandate

The technical co-founder departed after disagreement over using customer documents to improve retrieval and evaluation, six weeks before a contracted enterprise launch. The engineering team can finish features, but architectural decisions, security assurances and model-provider dependencies have no accountable executive owner.

The interim needs to commence within two weeks for eight months, covering launch, hypercare and the recruitment of a permanent CTO. Extension is possible for one month only if the selected successor cannot join before the post-launch assurance review.

Handover requires the enterprise customer to accept security, data isolation, retrieval quality and support criteria; the service must run within agreed reliability and inference-cost limits for ninety days; and the permanent CTO must lead one model-provider change exercise.

The interim may set architecture, stop launch, choose model-routing and allocate the approved ₹9 crore delivery budget. Use of customer data beyond contract, provider commitments above ₹10 crore, material scope concessions, permanent vice-president hires and acceptance of severe safety exceptions require CEO, customer or board consent.

Consumer product expansion, creation of proprietary foundation models and fundraising are excluded. The seat must deliver one secure enterprise product and an operable technology organisation, not pursue every possible AI research direction.

Why this seat is open

The co-founder exit removed the person carrying implicit architectural and customer assurances. Promoting a team lead now would preserve undocumented decisions and place them in a founder dispute. An interim CTO can convert those assumptions into explicit, testable obligations before permanent succession.

What you will own

• Decide the production architecture for retrieval, model routing, tenant isolation, audit logging and fallback operation.
• Convert enterprise security and data-use commitments into tested controls with retained customer acceptance evidence.
• Set model evaluation thresholds for groundedness, harmful output, leakage, latency and task completion.
• Approve launch only after load, red-team, rollback, support and inference-cost tests meet contracted conditions.
• Negotiate model-provider portability through abstraction, prompt and evaluation versioning, and exercised fallback.
• Govern ninety days of hypercare against reliability, quality, security and unit-cost measures.
• Transfer architecture records, customer obligations, model evaluations, provider economics and team decisions to the permanent CTO.

Candidate qualifications

• Served as CTO or engineering chief for enterprise AI, developer infrastructure or cloud-native software.
• Shipped a production generative-AI product using retrieval, model orchestration and strong tenant isolation.
• Negotiated enterprise security, data-use and acceptance conditions directly with customer technology leaders.
• Designed evaluations covering groundedness, leakage, safety, latency and business-task quality.
• Managed model-provider concentration and exercised fallback without losing audit lineage.
• Stabilised a technology organisation after founder or key-person departure.

Non-negotiables

• Can start in Hyderabad within fourteen days.
• No current financial interest in a foundation-model provider under evaluation.
• Will not use customer content for training or evaluation beyond explicit contract.
• Must be willing to delay launch if isolation, safety or rollback evidence fails.

1. 49 words maximum. State your availability and any foundation-model or customer conflict.
2. 49 words maximum. Describe a generative-AI enterprise launch you governed and its hardest acceptance test.
3. 49 words maximum. How have you proved model-provider fallback without losing output traceability?