AI Architect and AI Delivery Lead

Role Summary

The AI Architect and Delivery Lead is Bimeda's hands-on technical owner for AI. The role is responsible for being the translator of AI possibilities to reality. The person designs the architecture, helps and guides the builds, ships solutions, and ensures they carry into steady operation, so that Bimeda can build and run its own AI rather than rely on external partners for every build. The role leads any AI Developers hired and will serve as the AI tools’ Platform Administrator.

The role takes its priorities from the AI Council, and reports into IT for access, security, governance, and delivery standards. The person is accountable for the technical answer to how Bimeda builds and runs AI. It does not set AI policy, use-case priority, or spend, though it strongly influences all. Outputs must work, be documented, and be supportable by someone other than the person who built them.

About the Organization

Bimeda is a global animal health company operating in more than 80 countries, with manufacturing and commercial operations spanning North America, Latin America, Europe, and Africa/Middle East. Its AI programme runs in a GxP-regulated environment on Microsoft 365, Salesforce and QAD ERP, and has standardised on Claude as its primary AI productivity tool. The programme is steered by an Executive AI Subcommittee and delivered through a cross-functional AI Council that sets priorities across regions and functions.

Candidates should understand upfront that this is the first internal AI role at Bimeda, though others have been approved and will be hired. A share of the work is running, fixing, and improving solutions that already exist, including agents handed over by external partners, alongside new builds. Priorities can shift as the AI Council reprioritises across regions and business functions. Candidates should be prepared to explain how they would balance a new build against keeping live solutions healthy.

Ways of Working

This is a role for a candidate based in Brazil, working with colleagues in Dublin, and Chicago, and other main hubs around the world. A minimum of 3 working hours of daily overlap with Dublin and Chicago is expected. Successful candidates must be genuinely comfortable working independently so that they can guide their own technical approach, exercising judgment without consistent supervisor immediately available, and escalating only when a decision genuinely requires it.

The role holds privileged access to Bimeda's AI platform, connectors, and cloud environment. The role implements spend policy set by the AI Council and the Executive Committee.

Responsibilities

Architecture & Standards

- Own, document, and update the reference architecture and written decision records for how Bimeda builds and runs AI, including the technical recommendation between three delivery paths for each approved use case: buy off the shelf, build in-house, or engage a specialist partner.

- Set and maintain standards for agents, skills, and other AI utilities: logging, observability, evaluation, versioned prompts, and documented human-in-the-loop checkpoints.

- Give the AI Council a technical recommendation on each new use case at intake, with an effort and cost estimate and the data and security dependencies named.

Build & Delivery

- Build in-house agents and chatbots on the approved AI tool(s) and the Python scripting and integrations they need with Microsoft 365, Salesforce, the ERP, and any other core system.

- Be accountable for the build quality, testing, and documentation of in-house solutions, and direct the work of the AI Development team.

Operation of Live Solutions

- Act as technical owner of solutions built on Bimeda's AI platform, starting with agents handed over by an external partner, and as technical contact for vendor-run tools.

- Monitor, fix, and re-test live solutions, including when underlying AI models change, and keep run-books current.

Partner Oversight

- Review the technical deliverables in partner statements of work, accept or reject handovers, and flag where a build could be delivered in-house at lower cost.

- Support Bimeda's milestone- and deliverable-based approach to partner engagements; commercial terms remain with the VP Strategy & Transformation and VP IT.

Governance in Practice

- Turn AI policy and the governance roadmap into working controls: audit trails, access boundaries, and documented human decision points for any tool that supports a quality or GxP decision.

- Agree the validation approach with Quality before touching any tool that supports a quality or manufacturing decision (GxP).

Team Leadership & Enablement

- Lead the AI Development team: set priorities, review code and agent designs, and keep a documented backlog.

- Run short technical clinics for AI Council members and functional champions on what can and cannot be built, and what new AI functionality will change the organization.

AI Tool Fluency

- Advanced practical experience using the top AI tools (OpenAI, Anthropic, Google) a must. Claude specific experience strongly preferred. Experience with comparable AI tools as part of a regular engineering workflow, including building with them rather than only using them.

Other

Any such duties as may be required to support the effective functioning of the AI programme, commensurate with the grading of the post.

Qualification/Requirements

LANGUAGE SKILLS

- Fluency in English (both written and verbal).

- Additional languages are a plus with our core being Portuguese, Spanish, French.

EXPERIENCE

- 6+ years in software engineering or architecture roles, including 2+ years building production LLM applications or agents.

- Demonstrated experience shipping and operating an agent or automation used daily by non-technical colleagues. Candidates should be able to describe a specific instance where an architecture decision changed cost, risk or outcome, not only one that followed an existing plan.

- A genuine self-starter track record: examples of identifying a gap and building a solution without being directed to, not just executing well-defined tasks.

- Experience integrating with enterprise systems (Microsoft 365, Salesforce, or an ERP) and working cross-functionally with multiple stakeholders; comfortable with ambiguity. Prior experience supporting a multi-region or multinational organization is a strong plus.

- Experience in a regulated environment (pharmaceuticals, medical devices, finance), including exposure to computer system validation, is a plus.

- Able to read a partner statement of work and judge whether its technical deliverables are sound.

TECHNICAL

- Strong Python, plus cloud deployment and CI/CD (Azure preferred; AWS or GCP acceptable). This is the core technical requirement of the role.

- Hands-on use of the Claude API or SDK, or an equivalent: tool use, MCP connectors, multi-agent orchestration, and evaluating output quality.

- Written architecture decisions you can show, and working security habits: least-privilege access, secrets handling, audit logging, data protection.

- Low-code tools (Copilot Studio, n8n, Zapier) are a plus, not required, with judgment on when not to use them.

- Advanced technical degrees not a requirement. We are looking for doers.

- Strong documentation habits: run-books, reproducible methods, and version control.

BEHAVIOURAL

- Customer-focused; takes personal responsibility for speed, quality and reliability of delivery.

- Ability to work under pressure in a challenging and fast-moving environment.

- Excellent oral and written communication; able to explain a trade-off to a non-technical executive in one page and influence decisions through evidence.

Strong interpersonal skills; proactive, highly motivated self-starter; good judgment and discretion, including with privileged access. Genuinely comfortable operating independently with minimal day-to-day supervision (see Ways of Working above).