Manager - AI/ML (Hybrid)

Location: IN-KA-BENGALURU-NORTHGATE ~ Sy No 2/2 Venkatala Village ~ SY NO 2/2 VENKATALA VILLAGE, Yelahanka Hobli

 Position Role Type: Hybrid At RTX, the world's largest aerospace and defense company, 185,000 great minds are united by purpose and inspired to make a difference solving the world’s most complex problems. With our three market leading businesses, world-class operations and investments in research and development, we offer capabilities and opportunity no one else can. Together, we push the boundaries of known science and find new ways to connect and protect our world. Collins Aerospace is a leader in technologically advanced, intelligent solutions that help redefine the aerospace and defense industry. With a comprehensive portfolio and deep technical expertise, we help customers meet the demands of the global market. Join us and help shape the future of aerospace and defense. What You Will Do * Our Business Units bring us hard engineering problems — design cycles that take too long, certification evidence that is assembled by hand, simulation campaigns that cannot be run exhaustively, decades of legacy documentation nobody can search. The Applied Research & Technology (ART) organization exists to find out, quickly and honestly, which of those problems AI can actually solve. As an Engineering Manager you will lead one such team within the ART organization. You will run a portfolio of AI proof of concepts (PoCs) against BU-sponsored problems, prove or disprove value fast, and then make the call that matters most: hand the successful PoC to a central organization such as Digital Technologies (DT) to scale it into a deployable product at a Collins level. Leaving out the ones that do not work is as much a part of the job as scaling the ones that do.

1. Lead the team Line-manage and grow a team of 15–20 engineers spanning AI/ML engineering, software engineering, data engineering and domain-facing solution engineers, typically organized into 3–4 pods. Set technical direction, allocate people across concurrent PoCs, and protect focus when demand exceeds capacity. Own hiring, onboarding, performance management, career development and retention for the team. Build a culture where a disproven hypothesis delivered in six weeks is treated as a good outcome, not a failure. 2. Run the PoC portfolio Operate a structured intake process with Business Units: qualify incoming problems, challenge the framing, and convert vague asks into testable hypotheses with agreed success criteria before any code is written. Run PoCs on short, time-boxed cycles (typically 8–12 weeks) with stage gates, explicit kill criteria and a named BU sponsor for each. Maintain a live portfolio view — what is in flight, what it costs, what it is expected to return, and what has been stopped and why. Ensure every PoC produces a defensible evaluation: baseline comparison, measured performance against BU-relevant metrics, failure mode analysis, and an honest statement of what was not tested. 3. Decide and execute the path after PoC Own the transition recommendation for each successful PoC, backed by a written assessment of production readiness, data dependencies, integration surface, run cost and ownership model. Transition to DT or another central organization: package the PoC for handover — architecture documentation, model cards, evaluation results, data lineage, known limitations, backlog — and stay engaged through a defined support period so the receiving team is not left holding an artefact they cannot maintain. Scale in-house: where the capability is differentiating or too domain-specific to hand over, stand up the engineering to take it to deployment — CI/CD, MLOps, monitoring, model lifecycle management, user support and a funded product roadmap. Negotiate ownership, funding and SLA boundaries with DT & BU engineering leadership so that transitions are