Position Title

AI Architect (T & I) (Telework/Hybrid)

Employment Status

Permanent

Position Language Requirement

English, French

Language Skills

English (Reading - C - Advanced), English (Speaking - C - Advanced), English (Writing - B - Intermediate), French (Reading - C - Advanced), French (Speaking - C - Advanced), French (Writing - B - Intermediate)

Work Arrangements

This is a hybrid position with a mix of in‑office and remote work. Work arrangements will be discussed with hiring managers per departmental guidelines.

Unposting Date

:59 PM

Overview

Technology & Infrastructure (T&I;) is the backbone and future‑forward arm of CBC/Radio‑Canada. The team constantly innovates to evolve and maintain the Corporation’s technology and infrastructure, connecting media content, systems, people and places.

Your Role

As a key member of the Architecture and Strategic Development (Media System Architecture) team, you will design and plan CBC/Radio‑Canada’s future technology infrastructure. The mission is to transform disparate media workflows by integrating hybrid agents, video understanding systems and advanced artificial intelligence architectures operating 24/7.

Your Responsibilities

Optimizing Models, Inference and GPU Infrastructure

- Design and build high‑performance training and inference systems for LLMs and multimodal AI models.
- Optimize end‑to‑end training, including high‑throughput data pipelines (streaming, sharding, bucketing), distributed training (parallelism strategies) and mixed precision.
- Optimize inference and serving engines to guarantee minimum latency using KV caching design/management, batching, quantization and long‑context processing.
- Collaborate with the T&I; team to design, right‑size and evolve our internal GPU cluster into a current‑generation infrastructure capable of supporting these massive workloads.

Strategy, Architecture and Media Ecosystem

- Plan, develop and architect the integration of AI solutions within CBC/Radio‑Canada’s media production environments (PAM, MAM, television and radio studios).
- Evaluate and standardize integration protocols such as Model Context Protocol (MCP) and agent‑to‑agent (A2A)



architectures to interconnect our models with complex production platforms (Avid, Dalet, Adobe).
- Ensure absolute data security while developing video understanding systems that can interact with both file‑based workflows (codecs, wrappers) and IP live workflows (SMPTE‑2110).

Technical Leadership, Mentoring and MLOps

- Provide expert guidance on system efficiency, spearheading initiatives to dramatically improve architectural scalability, latency and reliability.
- Mentor and elevate the organization’s engineers and data scientists in large‑scale ML system design and performance engineering.
- Develop robust MLOps/LLMOps pipelines with a constant focus on observability, performance profiling and automated testing.
- Summarize and explain technological innovations to make them understandable to decision‑makers and production teams and influence our strategic investments.

Featured Technology Environment

- Inference Modelling and Optimization: PyTorch, vLLM, DeepSpeed, Accelerate, KV caching architectures, quantization techniques.
- Infrastructure, Cloud and Compute: internal GPU clusters, distributed architectures, AWS, Azure, GCP, containers, Kubernetes.
- Orchestration and Agents: Open WebUI, integration protocols (MCP, A2A), LiteLLM, LangChain, LangGraph, Strands.
- Media Production: SMPTE‑2110 protocols, PAM/MAM (Avid Interplay, Dalet), Adobe suite, media processing (audio/video codecs).
- MLOps and Observability: Langfuse, Weights & Biases (W&B;), MLflow, performance profiling tools.

Minimum Qualifications

- Bachelor’s or master’s degree in software engineering, information technology, artificial intelligence, mathematics or a related natural science field.
- Functional bilingualism (English and French) essential for Canada‑wide communications.
- At least five years’ proven experience developing and deploying AI/ML solutions.




- At least eight years’ experience building tools and platforms in a software engineering role.
- Demonstrated experience working with language models and designing solutions optimized for cost efficiency and scale.
- Strong conceptual understanding of LLM, RAG and AI agent architectures, including their frameworks and operational constraints.
- Experience selecting AI framework architectures (e.g., TensorFlow, PyTorch, Hugging Face), cloud platforms (Azure, AWS, GCP) and orchestration tools (Docker, Kubernetes) for scalable enterprise AI solutions.
- Knowledge of ModelOps, AI engineering, DevOps and MLOps practices (including CI/CD pipelines).
- Solid understanding of machine learning and deep learning fundamentals.
- Strong technical documentation skills, with the ability to produce diagrams, demos and technical artifacts that make AI architectures understandable and actionable.
- Hands‑on technical experience working with media production platforms (MAM/PAM) and designing scalable solutions in a highly available, 24/7 environment.
- Solid working knowledge of cloud technologies (AWS, Azure or GCP), virtualization, networking and storage.

Nice‑to‑Have Skills (Assets)

- Experience with modern video software architectures (e.g., NVIDIA Holoscan for Media) or ultra‑high‑performance IP transfer protocols.
- Demonstrated experience managing complex workflows through cloud or hybrid data pipelines.
- Active contributions to open‑source high‑performance AI projects and published research, or deep understanding of security frameworks for responsible AI.

Candidate Assessment

Candidates may be subject to skills and knowledge testing.

Background Check

- A mandatory criminal record check.
- Other background checks may be conducted based on the operational requirements of the position.

Equality and Diversity

CBC/Radio‑Canada is committed to being a leader in reflecting our country’s diversity. We value equal opportunity and nurture an inclusive workplace where our individual differences are recognized and valued.

Primary Location

1000, Rue Papineau, Montreal, Quebec, H2K 0C2

Number Of Openings

1

Work Schedule

Full time

#J-18808-Ljbffr

📌 AI Architect (T & I) (Telework/Hybrid)
🏢 Radio-Canada
📍 Montreal