About Johnson Controls

Johnson Controls, a global leader in thermal management, mission-critical building systems, energy efficiency, and decarbonization, helps customers use energy more productively, reduce carbon emissions, and operate with the precision and resilience required in rapidly expanding industries such as data centers, healthcare, pharmaceuticals, advanced manufacturing, and higher education.

For more than 140 years, Johnson Controls has delivered performance where it really matters. Backed by advanced technology, lifecycle services and an industry-leading field organization, we elevate customer performance, turn goals into real-world results and help move society forward.

Visit johnsoncontrols.com for more information and follow @Johnsoncontrols on social platforms. 

What you will do

Johnson Controls builds configured air handling units. A consulting engineer's mechanical schedule becomes a selection, a quote, a released BOM, a routing, and a build packet at a station. Along that path we generate a large amount of structured and semi-structured data — decades of historical orders, configurations, BOMs, labor actuals, test results, quality findings, and service records — and we currently use very little of it to make the next order faster, cheaper, or more accurate.

As a Senior AI Engineer, you will set the technical direction for AI at that intersection. Concretely: reading customer specifications and schedules well enough to propose a configuration, catching a wrong BOM before material is committed, estimating cost and labor for non-standard scope, and putting useful assistance in front of a sales engineer, an application engineer, and an operator on the floor. You will write production code, own architecture, and be the person the organization trusts to say which problems are genuinely AI problems, which are data quality problems wearing an AI costume, and which are better solved with a rule, a query, or a fixed process. We would rather hire someone with the judgment to challenge a bad idea early than someone eager to ship a demo.

Join our team in York, Pennsylvania, where you'll have the opportunity to make an immediate impact. Candidates should be located within commuting distance of the facility or be open to relocating, with relocation assistance available. For the right candidate, we are also open to considering a fully remote work arrangement.

How you will do it

Set technical direction

Own the AI/ML architecture and technical strategy for sales, engineering, and manufacturing systems — model selection, data architecture, evaluation approach, deployment topology, and build-versus-buy recommendations.
Partner with product management, sales operations, engineering leadership, and plant operations to translate business problems into a prioritized portfolio of AI work with honest, expected value and honest feasibility.
Establish the engineering standards this domain requires: reproducible training and evaluation, model and prompt versioning, monitoring and drift detection, rollback paths, and audit trails sufficient for a system that influences what gets built and shipped.
Serve as the senior technical voice in design reviews and as a mentor to software and data engineers adopting these techniques. Raise the team's ceiling, not just your own output.

Build the applications

Specification and document intelligence. Extract structured requirements from customer specifications, mechanical schedules, and submittal documents — PDFs, drawings, spreadsheets — and map them to candidate configurations and selections.
Configuration and quoting assistance. Guided selling and configuration recommendation, competitive crossover matching, pricing and discount guidance, and cost estimation for the non-standard and special scope that drives disproportionate engineering hours and margin risk.
Engineering automation. Similar-order retrieval and reuse, BOM anomaly detection that flags a released structure deviating from statistically comparable units, configuration rule mining and conflict detection over historical orders, and engineering change impact analysis.
Manufacturing intelligence. Labor hour and routing time prediction from configuration features, material shortage and schedule risk forecasting, anomaly detection over final run and leak test data, and quality defect prediction. Where it earns its place, computer vision for assembly verification, nameplate and label validation, or completeness checks at station.
Assistive interfaces. Retrieval-grounded assistants over product, application engineering, submittal, and service knowledge for sales engineers, application engineers, and service technicians — including agents that call configurator, PLM, and ERP APIs to do real work rather than just answer questions.

Make it safe to deploy

Design for the failure mode that matters here: a plausible-looking wrong answer that reaches production. AI advises; systems of record decide. Human-in-the-loop review, confidence thresholds, graceful abstention, and full traceability are default requirements, not enhancements.
Build evaluation before building features. Define ground truth with the engineers and estimators who own the judgment today, and measure against it continuously in production.
Solve for real deployment constraints — on-premise and edge inference at plants, latency budgets at a build station, intermittent connectivity, OT/IT network segmentation, and plant cybersecurity requirements.
Establish data governance for AI: what customer, pricing, and design data may leave our environment, what may be sent to third-party model providers, retention and tenancy, and how vendor terms are evaluated.

Work the data foundation

Assess and improve the usability of source data across [CPQ], [PLM], [ERP], and [MES]. Be the person who says plainly when a model cannot succeed until master data, part number governance, or labor reporting accuracy improves — and who then helps drive that fix.
Build the pipelines, feature engineering, and datasets that make repeated AI development cheap rather than one-off.

Partner on delivery

Work in close partnership with the Program Manager for these systems. They own scope, schedule, and stakeholder accountability; you own technical direction and feasibility. Neither role works without candid, early information from the other about what is actually achievable and when.

What you will need
Required

Bachelor's degree in Computer Science, Engineering, Applied Mathematics, Statistics, or a related field.
5+ years in software or data engineering, with 5+ years building machine learning or AI systems that reached production and real users.
Demonstrated ownership of an AI/ML system in production over time — not proofs of concept. You should be able to describe what broke after launch, how you found out, and what you changed.
Strong Python and SQL. Depth in the modern applied stack: classical and tabular ML (gradient boosting, scikit-learn), deep learning frameworks, and LLM application patterns including retrieval, structured extraction, tool use, and evaluation.
Real fluency with messy enterprise data — ERP and PLM schemas, MES and historian data, semi-structured documents — and the practical experience to know that this is where most of the work lives.
Experience deploying and operating models in production: containerization, CI/CD, cloud ML platforms, monitoring, and cost management.

Preferred

Experience in manufacturing, industrial, or engineer-to-order environments — HVAC, capital equipment, building products, or similar highly configured products.
Familiarity with the configure-to-order data landscape: BOM structures, variant and option modeling, routings and standard times, engineering change processes.
Document AI and information extraction experience with engineering drawings, specifications, or technical documentation.
Optimization and operations research exposure — scheduling, sequencing, nesting, capacity planning — and the judgment to reach for it instead of ML when it is the right tool.
Computer vision experience in an industrial setting, including the practical realities of lighting, fixturing, and line-side integration.
Edge or on-premise inference experience in an OT environment.
Experience being the first or one of the first senior AI hires somewhere, building the function rather than inheriting it.

HIRING SALARY RANGE: $120,000 - 155,000 (Salary to be determined by the education, experience, knowledge, skills, and abilities of the applicant, internal equity, and alignment with market data.) This role offers a competitive Bonus plan that will take into account individual, group, and corporate performance. This position includes a competitive benefits package. The posted salary range reflects the target compensation for this role. However, we recognize that exceptional candidates may bring unique skills and experiences that exceed the typical profile. If you believe your background warrants consideration beyond the stated range, we encourage you to apply. To support an efficient and fair hiring process, we may use technology assisted tools, including artificial intelligence (AI), to help identify and evaluate candidates. All hiring decisions are ultimately made by human reviewers. For details, please visit the About Us tab on the Johnson Controls Careers site at https://jobs.johnsoncontrols.com/about-us

#LI-HYBRID

#LI-REMOTE

Johnson Controls International plc. is an equal employment opportunity and affirmative action employer and all qualified applicants will receive consideration for employment without regard to race, color, religion, sex, national origin, age, protected veteran status, genetic information, sexual orientation, gender identity, status as a qualified individual with a disability or any other characteristic protected by law. To view more information about your equal opportunity and non-discrimination rights as a candidate, visit EEO is the Law. If you are an individual with a disability and you require an accommodation during the application process, please visit here.

Senior AI Engineer (hybrid)

About the Role

Skills

Benefits

Perks

Full job details