ML Solutions Architect

About the role

NBCUniversal is looking for an ML Solutions Architect in Montreal, a systems-minded software engineer with strong general software engineering expertise and solid machine-learning adjacent experience. You will lead the high-level design of systems that integrate ML models into the company's product suite and act as a technical consultant, deciding whether customer needs can be met with off-the-shelf solutions or should be escalated to the Deep Learning and Reinforcement Learning teams.

Key responsibilities

• Design and implement the software layers that let ML models interact with a real-time rendering engine, including data pre- and post-processing (coprocessing).
• Evaluate customer requirements and choose between off-the-shelf tools and specialized research projects for the Deep Learning or Reinforcement Learning teams.
• Build and maintain language-agnostic wrappers, APIs and microservices so the ML stack works across production environments.
• Act as the main technical liaison between leadership, customers and the engineering team to define data and integration requirements.
• Break complex product visions into modular architectural components so ML features ship within a stable, scalable product.

Required profile

• Degree in Computer Science, Software Engineering or a related field.
• Proven experience as a Software Architect, Solutions Architect or Systems Engineer in a fast-paced environment.
• Prior experience in complex, multi-disciplinary fields such as robotics, smart grids, precision agriculture, game development or aerospace is strongly valued.
• Ability to run build-vs-buy analyses for ML components and to turn product vision into concrete engineering specs.

Required skills

• Fluent use of Git and the Unix shell, working across several languages (ideally Python, C++ or C#).
• Deep knowledge of integrating ML models into production software: API design, message brokers, containerization, and compute and memory budgeting.
• Fluency in 3D geometry, linear algebra and/or computer graphics.
• Enough ML-adjacent experience to understand model constraints, data requirements and the state of the art; fine-tuning and deployment experience is a plus.

The position is hybrid, with a minimum of four days per week in the office, and external candidates may be asked to attend an in-person interview before a hiring decision.