Director, Commercial AI Orchestration Architect

Bristol Myers Squibb•Princeton, NJ

1d•Remote

About The Position

Bristol Myers Squibb is making a strategic shift from traditional next-best-action solutions toward an enterprise agentic orchestration capability that can dynamically coordinate personalized HCP engagement across field and non-personal channels. We are building a new orchestration foundation that brings together customer signals, commercial priorities, approved content, channel eligibility, and business guardrails to generate and execute coordinated customer journeys. The ambition is not to create another stand-alone recommendation engine, but to establish a reusable capability that can continuously determine the most appropriate sequence of engagements, adapt based on new signals and outcomes, and improve how BMS engages customers across channels. We are looking for a Director, Agentic Commercial Orchestration Architect to serve as BMS’s technical authority for this strategic program. This individual will own the end-to-end orchestration architecture, shape the technical direction of the co-build with an implementation partner, and ensure the capability is scalable, compliant, explainable, and ultimately owned and operated by Bristol Myers Squibb. The role will define how agents, decisioning models, optimization methods, commercial rules, customer context, and channel-execution services work together as one orchestration capability. It will also establish the architecture, engineering standards, evaluation framework, and technical guardrails required to move from an initial use case to a sustainable enterprise capability. Success means more than delivering the first implementation. It means creating an orchestration foundation that Bristol Myers Squibb can independently operate, extend across brands and channels, and continuously improve without becoming dependent on a single implementation partner, AI model, or proprietary orchestration framework.

Requirements

Proven experience designing and delivering production-grade next-best-action, next-best-engagement, journey optimization, decisioning, or omnichannel orchestration platforms.
Strong understanding of multi-agent architecture, including agent decomposition, orchestration patterns, shared state, context management, agent-to-agent handoffs, tool selection, failure handling, and human-in-the-loop controls.
Hands-on experience with agent frameworks such as LangGraph, Semantic Kernel, AutoGen, LangChain, or comparable orchestration frameworks.
Experience designing agent tool access using Model Context Protocol (MCP), APIs, event-driven integration, webhooks, and asynchronous messaging patterns.
Strong knowledge of context engineering, prompt management, structured outputs, tool calling, grounding, retrieval-augmented generation, and agent memory patterns.
Experience defining and managing customer and journey state across multiple agents, interactions, channels, and execution cycles.
Ability to design rules engines, constraint-based decisioning, prioritization logic, optimization models, and AI-driven decisioning within a common orchestration architecture.
Understanding of decisioning methods such as contextual bandits, reinforcement learning, predictive scoring, mathematical optimization, and hybrid rules-and-model approaches.
Strong understanding of closed-loop orchestration, including execution acknowledgements, response signals, outcome measurement, feedback loops, model evaluation, and continuous optimization.
Experience designing explainable and auditable decisioning, including decision traces, reason codes, model and prompt versioning, input lineage, configuration history, and reproducibility.
Knowledge of agent evaluation frameworks, including offline evaluation, simulation, regression testing, adversarial testing, guardrail testing, journey-quality measures, and production monitoring.
Experience establishing LLMOps and MLOps practices, including CI/CD, model and prompt lifecycle management, evaluation gates, environment promotion, observability, rollback, and incident management.
Strong understanding of production AI concerns such as latency, token usage, cost controls, caching, concurrency, rate limits, retries, idempotency, circuit breakers, and graceful degradation.
Experience integrating orchestration platforms with CRM, commercial data products, content-management platforms, consent services, identity systems, and channel-execution platforms.
Experience designing secure enterprise integrations using OAuth 2.0, workload identity, role-based access control, secrets management, encryption, API gateways, and complete audit logging.
Ability to define canonical data contracts and semantic models for HCP context, customer signals, barriers, engagement history, recommendations, journeys, content, channel eligibility, and execution outcomes.
Experience with configuration-driven architecture that allows business rules, brand guardrails, channel constraints, and market-specific requirements to evolve without hard-coding them into individual agents.
We hire for skills and capabilities, not just credentials – if this role excites you, but doesn’t perfectly match your resume, we encourage you to apply anyway.

Responsibilities

Own the end-to-end architecture of the agentic orchestration capability, including experience composition, cross-channel optimization, journey generation, execution, feedback, and continuous learning.
Define how orchestration consumes customer signals, barrier insights, brand priorities, approved content, channel eligibility, consent, suppression rules, and engagement history.
Establish clear boundaries and interfaces between insight generation, decisioning, journey orchestration, channel execution, and outcome measurement.
Define the architecture of the agents, their responsibilities, tool access, decision rights, escalation paths, and interaction patterns.
Design the orchestration solution to support field calls, remote engagement, approved email, events, headquarters email, and third-party channels.
Translate commercial objectives and brand guardrails into configurable orchestration rules and testable agent behaviors.
Define technical acceptance criteria and the definition of done for all major engineering deliverables.
Validate critical design decisions through code reviews, prototypes, technical walkthroughs, test evidence, and production-readiness assessments.
Ensure every recommendation and action is explainable, traceable, compliant, and reproducible.
Establish evaluation methods for journey quality, recommendation relevance, guardrail adherence, execution success, and measurable business outcomes.
Define the architecture for human review, exception handling, approval, escalation, and safe fallback when an agent cannot act confidently.
Ensure orchestration logic, source code, configurations, documentation, and operational knowledge are transferred to and owned by BMS.
Build a sustainable, BMS-owned orchestration capability by establishing engineering standards and enabling hands-on knowledge transfer with the implementation partner, while maintaining architectural portability and avoiding dependency on any single partner, AI model, or proprietary orchestration framework.

Benefits

wellbeing support
retirement and financial protection benefits
insurance offerings (medical, dental, vision, life and disability)
Flexible Time Off (FTO)
11 paid company holidays each year
160 hours of paid vacation annually for new hires
3 optional holidays
paid sick leave
up to two paid volunteer days per year
summer hours flexibility
leaves of absence for medical, personal, parental, caregiver, bereavement, or military needs
annual Global Shutdown between Christmas Day and New Year's Day