Director, Applied Artificial Intelligence

TraceLink•Wilmington, MA

19h

About The Position

TraceLink is seeking a Director of Applied Artificial Intelligence to lead the team responsible for designing, building, evaluating, and operating the cognitive functions that drive agentic orchestration for every OPUS Agent. This role is crucial for developing the 'OPUS Brain,' the cognitive layer that enables AI agents to perceive, reason, plan, decide, coordinate, and learn within defined governance boundaries. The position requires hands-on leadership at the intersection of AI research, cognitive architecture, and supply chain operations, focusing on integrating various AI technologies into a reusable cognitive architecture that serves multiple agent roles across multienterprise processes. A key aspect of this role is ensuring that the OPUS Brain enforces governance by design, keeping agents within their scoped permissions, rules, and objectives, and ensuring accountability to human managers.

Requirements

10+ years of experience, including 4+ years in machine learning, AI, or data science.
3+ years leading applied AI or ML engineering teams.
A track record of shipping LLM-based or agentic systems into production for enterprise customers.
Deep hands-on expertise in agent architectures, including LLM reasoning, planning and task decomposition, tool use, agent memory, retrieval-augmented generation, and multi-agent orchestration.
Experience designing evaluation methods for agent reasoning, planning, and decision quality, not only model outputs.
Experience building AI systems with identity, role-based access, permissions, guardrails, and full audit trails.
Strong software engineering foundations, including cloud platforms (AWS, Azure, or GCP), APIs, event-driven architectures, and distributed systems.
Ability to explain complex AI concepts, including how agents reason and decide, to executives, customers, and non-technical operators.
Bachelor's degree in Computer Science, Engineering, Mathematics, or a related field.

Nice To Haves

Master's or PhD highly desired.
Background in cognitive architectures, automated planning, symbolic or neuro-symbolic reasoning, or decision theory.
Experience with knowledge graphs, digital twins, optimization, or simulation environments for agent testing.
Domain experience in supply chain, logistics, manufacturing, or life sciences operations.
Familiarity with multienterprise networks, B2B integration, EDI, or supply chain control towers.
Experience in regulated environments such as FDA, EU GMP, or DSCSA compliance.
Published research, patents, or open-source contributions in agentic systems, reasoning, or planning.

Responsibilities

Define and own the architecture and multi-year roadmap for the OPUS Brain as the shared cognitive foundation for all OPUS Agents and Agentic Control Towers.
Establish a modular set of cognitive functions (perception, memory, reasoning, planning, decisioning, orchestration, reflection) that agent roles can compose and configure.
Evolve the OPUS Brain's capabilities to enable agents to progress from analysis and recommendation to governed autonomous execution.
Make build-versus-buy and model-selection decisions for foundation models, reasoning models, retrieval systems, and classical ML components.
Ensure the OPUS Brain enforces OPUS Agent Profiles, roles, permissions, objectives, Tasks, Decisions, and Rules at every step of reasoning and action.
Make agent reasoning traceable by linking every Decision to the context, knowledge, and Rules that produced it and capturing this in the audit log.
Define human-in-the-loop patterns within the cognitive architecture, including confidence thresholds, escalation triggers, and approval checkpoints.
Protect against failure modes such as hallucination, goal drift, permission overreach, and unsafe tool use through guardrails and runtime controls.
Build rigorous evaluation frameworks for each cognitive function and for end-to-end orchestration, covering reasoning quality, plan correctness, decision accuracy, adherence to Rules and permissions, reliability, cost, and latency.
Create simulation and scenario-testing environments, grounded in the Digital Twin, for testing agents against realistic supply chain disruptions.
Own observability of agent reasoning, drift detection, incident response, and continuous improvement.
Partner with Security, Compliance, and Legal to meet regulatory requirements in highly regulated industries.
Define responsible AI standards for transparency, explainability, data privacy, and auditability of agent reasoning and decisions.
Make the OPUS Brain's governance a product feature that customers can inspect and trust.
Hire, develop, and lead a high-performing team spanning applied AI research, ML engineering, and agent systems.
Set a culture of shipping, measurement, and accountability within the team.
Work with Product and Platform Engineering to enable agent teams to configure OPUS Brain capabilities for new operational roles quickly and safely.
Work with Customer Success to bring agents into customer operations and translate feedback from human managers into improvements in the cognitive architecture.
Represent TraceLink's work on agentic cognition with customers, partners, analysts, and industry forums.