Do the AI work that reaches production.

About the role

Most AI programs fail before a line of code is written: the wrong use case, an unmeasured value case, or a pilot nobody can explain to finance. The AI Product Manager sits with the client's business owners, turns ambition into a ranked backlog of use cases with a value hypothesis each, and holds the line on scope once building starts. You are the person who can say "this one does not pay" and make it stick.

What you will do

• Run discovery with business stakeholders to surface, size and rank candidate AI use cases against value, feasibility and risk.
• Write the value case for each use case: baseline, target metric, cost to build and run, and the decision it informs.
• Own the roadmap and backlog for the engagement, and sequence work with the AI Solutions Architect and delivery pod.
• Define success criteria and evaluation targets before build starts, and report against them weekly.
• Facilitate pilot reviews and go or no-go decisions, and retire use cases that miss their targets.
• Translate between executives, business users and engineers so every group hears the same plan.

What you bring

• Five or more years in product management, with at least one product that shipped and was measured against a business outcome.
• Hands-on familiarity with LLM-based products: what retrieval, agents and evaluation actually mean in practice.
• Comfort with numbers: building a value case in a spreadsheet, reading an eval dashboard, challenging a cost estimate.
• A record of saying no to stakeholders with data, and keeping the relationship.
• Clear written communication. Your one-page use-case briefs are the artifact the whole engagement runs on.
• Experience in at least one of our core sectors: financial services, healthcare, manufacturing, public sector or B2B software.

Nice to have

• Prior consulting or client-facing delivery experience.
• Exposure to AI risk classification and governance requirements.

Apply for this role

About the role

The AI Solutions Architect designs the system around the model. Retrieval, orchestration, evaluation, guardrails, cost and the integration with the client's systems of record are your decisions, made early and written down. You are usually the first specialist on an engagement after the use case is chosen, and the one who can say what will break at ten times the load.

What you will do

• Produce the target architecture for each use case: model and vendor choices, retrieval design, orchestration pattern, evaluation plan and cost model.
• Review existing pilots and write a findings memo the client can act on: what scales, what does not, what to change first.
• Set engineering standards for the engagement, including prompt and model versioning, evaluation gates and observability.
• Make build-versus-buy calls with evidence, and defend them to the client's architecture board.
• Pair with AI Engineers and Data Engineers on the riskiest path through the system, and unblock them when the design meets reality.
• Work with the AI Security Engineer and AI Governance Advisor so controls are designed in, not bolted on.

What you bring

• Ten or more years in software or data engineering, including at least three years designing and shipping LLM-based systems in production.
• Deep working knowledge of at least one major cloud AI stack: Azure OpenAI, AWS Bedrock or Google Vertex, plus the surrounding services.
• Practical experience with retrieval architectures, vector stores, agent frameworks and evaluation harnesses.
• The ability to estimate cost and latency at scale and to explain the trade-offs to a non-technical executive.
• Strong written architecture documentation. Decisions you make are read by people who were not in the room.
• Experience integrating with enterprise systems: identity, data platforms, ERP or core systems in a regulated industry.

Nice to have

• Cloud architecture certifications from a major provider.
• Experience with on-premises or hybrid deployments and open-weight models.

Apply for this role

About the role

Our AI Engineers are forward-deployed. You shadow the client's business process before writing code, put a working prototype in front of real users within the first two weeks, and ship version one to a pilot group within the first month. Then you harden it: evaluations, cost, latency, error handling and the integration glue that turns a demo into a system people rely on. You write the evals before the prompt.

What you will do

• Build LLM-powered features end to end: retrieval pipelines, agent workflows, tool integrations, APIs and user-facing surfaces.
• Define and maintain evaluation sets for every feature you own, and report quality, latency and cost baselines.
• Work directly with business users to discover requirements, run weekly feedback loops and adjust the product to how work actually happens.
• Integrate with the client's systems: authentication, permissions, data access, logging and observability in their environment.
• Hand over working software with documentation and a client engineer who can run it.
• Lead a delivery pod when the engagement calls for it, coordinating a Data Engineer, LLMOps Engineer and support roles.

What you bring

• Five or more years of software engineering, with at least one LLM-based feature shipped to production and used by real customers or staff.
• Strong Python, and comfort in TypeScript or another production language the client uses.
• Hands-on experience with model APIs from OpenAI, Anthropic or an open-weight provider, and with frameworks such as LangGraph or Semantic Kernel.
• A habit of measuring: you can show an eval dashboard for something you built.
• Customer-facing communication. You can explain a trade-off to an operations manager and take feedback without defensiveness.
• Comfort with ambiguity and changing priorities on a client floor.

Nice to have

• Voice, real-time or multimodal experience.
• Prior consulting, solutions engineering or forward-deployed experience.

Apply for this role

About the role

Models are only as useful as the data they can reach, and in most enterprises that data is locked in systems of record with their own permissions, refresh cycles and quality problems. The Data Engineer makes it usable: ingestion, chunking, embeddings, lineage, freshness and access control that respects who is allowed to see what. When retrieval quality is measured, your pipeline is what is being measured.

What you will do

• Inventory the data an AI use case needs: sources, owners, sensitivity, refresh cadence and access rules.
• Build governed pipelines from systems of record into vector stores and feature layers, with lineage and monitoring.
• Design chunking, embedding and metadata strategies and measure their effect on retrieval quality against labeled sets.
• Implement document-level and row-level permissions so retrieval never returns content a user cannot see.
• Set freshness targets and alerting, and own the fix when a source changes shape.
• Work with the AI Governance Advisor on data classification and with the AI Security Engineer on exfiltration risk.

What you bring

• Five or more years of data engineering in production, with hands-on work on unstructured and vector pipelines.
• Strong SQL and Python, and experience with at least one modern platform: Snowflake, Databricks or BigQuery.
• Orchestration and transformation tooling such as Airflow and dbt, and streaming experience with Kafka or an equivalent.
• Familiarity with vector stores such as pgvector, Pinecone or Weaviate, and with embedding models and their trade-offs.
• A working understanding of data governance: catalogs, lineage, classification and access control.
• Experience with enterprise sources: document management, CRM, ERP or clinical and financial systems.

Nice to have

• Unity Catalog, Purview or another enterprise governance layer.
• Experience in a regulated industry with audit requirements.

Apply for this role

About the role

The LLMOps Engineer owns everything that happens after a feature is "done" and has to run reliably at scale. Prompt and model versions, evaluation gates in the release pipeline, tracing of every interaction, cost dashboards, provider routing and the incident response when a model starts answering differently in a regulated context. You apply DevOps discipline to systems that are stochastic by nature.

What you will do

• Build and operate the deployment pipeline for prompts, models and retrieval configurations, with versioning and fast rollback.
• Maintain the evaluation harness that every change must pass, covering quality, safety, format and regression checks.
• Instrument LLM interactions end to end: prompts, retrieved context, responses, latency, token cost and errors.
• Design model routing and fallback across providers, and manage rate limits, quotas and caching.
• Own cost and latency budgets, and report against them to the client and the Norton engagement lead.
• Run incident response for AI-specific failures, including quality drift, and feed findings back into the eval set.

What you bring

• Five or more years in DevOps, MLOps or platform engineering, with production experience operating LLM-based services.
• Strong Python and infrastructure-as-code, and fluency with containers, Kubernetes and CI/CD.
• Hands-on experience with LLM observability and evaluation tooling, and with building an eval harness from scratch.
• Practical knowledge of cloud AI services and gateway patterns on AWS, Azure or Google Cloud.
• Comfort reading cost and latency data and making routing or caching decisions from it.
• Experience running on-call and writing post-incident reviews people actually read.

Nice to have

• Experience serving open-weight models with vLLM or similar.
• FinOps experience or a track record of reducing inference spend.

Apply for this role

About the role

LLM systems have an attack surface traditional application security was not built for: instructions hidden in retrieved documents, agents with tool access they should not have, and models that leak what they were told. The AI Security Engineer threat-models each use case, tests it adversarially before and after launch, and designs the controls that let a regulated client sign off. You work from the OWASP Top 10 for LLM applications as a floor, not a ceiling.

What you will do

• Threat-model AI use cases across the alignment, instruction-following and tool-call layers, and document the risks in language the client's security team recognizes.
• Run structured red-team exercises for prompt injection, jailbreaks, indirect injection through retrieved content and data exfiltration, using automated tooling plus manual testing.
• Design agent permission scoping, least-privilege tool access and human-in-the-loop controls for high-impact actions.
• Build guardrails, input and output filtering and monitoring that feed the LLMOps evaluation harness.
• Review vendor and model supply-chain risk, and support client security reviews and penetration-test findings.
• Train client engineers on secure patterns so the controls survive after the engagement ends.

What you bring

• Five or more years in application security, penetration testing or security engineering, with hands-on work on LLM or agent-based systems.
• Working knowledge of the OWASP Top 10 for LLM applications and of current prompt-injection and jailbreak techniques.
• Experience with red-teaming tooling for AI systems and with writing findings that lead to fixes.
• Solid software skills in Python and an understanding of how retrieval, agents and tool calling are implemented.
• Familiarity with identity, secrets and access-control patterns in at least one major cloud.
• The ability to brief a CISO and pair with an engineer in the same afternoon.

Nice to have

• Security certifications such as OSCP, CISSP or cloud security specialties.
• Experience supporting SOC 2, HIPAA or financial-services security reviews.

Apply for this role

About the role

The AI Governance Advisor sits between three groups: engineers who need a concrete checklist, lawyers and risk officers who need evidence the checklist was followed, and executives who need a plain answer on exposure. You build the model inventory, classify each system by risk, write the policies that fit the client's obligations, and make sure the audit trail exists before anyone asks for it.

What you will do

• Establish and maintain the client's AI system inventory: purpose, data inputs, decision impact, owners and vendors.
• Classify use cases by risk tier against the frameworks that apply to the client, such as the NIST AI Risk Management Framework, ISO/IEC 42001 and the EU AI Act where relevant.
• Draft and operationalize AI policy: acceptable use, human oversight, data handling, vendor review and incident reporting.
• Run impact assessments for privacy, fairness and regulatory exposure, and document them in a form a regulator or auditor would accept.
• Design the audit trail with the LLMOps Engineer: what is logged, retained and reviewable, and by whom.
• Brief executives, legal and risk committees, and keep approvals moving so delivery is not blocked.

What you bring

• Seven or more years in risk, compliance, privacy, legal operations or technology governance, with direct AI or model-risk experience.
• Working knowledge of the NIST AI RMF and ISO/IEC 42001, and familiarity with sector regulation in at least one of financial services, healthcare or public sector.
• Enough technical literacy to read an architecture diagram and ask the right question about data flow and retention.
• A record of writing policy that engineers can actually follow and auditors can actually verify.
• Facilitation and executive communication skills. You run the meeting where the decision gets made.
• Experience with vendor and third-party risk assessment.

Nice to have

• Certifications such as AIGP, CIPP or CRISC.
• Model risk management experience under SR 11-7 or equivalent.

Apply for this role

About the role

A working AI system that nobody uses is a failed project with good engineering. The AI Enablement Lead owns adoption as a measurable outcome: how work changes, who champions it, what training gets people productive and whether the numbers move. This is a change-and-adoption role with real accountability, not a communications function.

What you will do

• Build the adoption plan for each use case: target users, workflow changes, champions, training and the metrics that prove it worked.
• Design and deliver hands-on training, role-specific playbooks and onboarding for business users and managers.
• Recruit and coach a champion network inside the client, and run the feedback loop back to the delivery pod.
• Redesign workflows with process owners so the AI system fits how work is done, and retire the old path.
• Instrument and report adoption: active use, task completion, time saved, quality outcomes and user sentiment.
• Partner with the AI Governance Advisor on responsible-use guidance that people can follow.

What you bring

• Seven or more years in change management, learning and development, operations or product adoption, with at least one technology rollout you owned end to end.
• Hands-on fluency with current AI tools and a clear view of what they can and cannot do for a business user.
• Facilitation and training design skills, from a 20-minute floor briefing to a multi-week learning path.
• Comfort with adoption analytics: defining a metric, instrumenting it and presenting it to a skeptical executive.
• Credibility with frontline teams and managers alike, and the patience to work through resistance.
• Strong writing. Your playbooks are how the client remembers what good looks like.

Nice to have

• Prosci or equivalent change-management certification.
• Experience in contact centers, claims, clinical operations or another high-volume workflow.

Apply for this role