Six modules from inference foundations to production certification

Structured learning paths for ML engineers, DevOps leads and platform architects building real-time AI infrastructure in Canada. Each module combines hands-on lab work with production-grade patterns — not generic online courses or IT training unrelated to model serving.

Inference lab session at InferenceAI Toronto studio

Lab-first delivery

Every programme module runs through our Eglinton corridor studio or remote lab environments with live GPU access. Participants work on actual inference pipelines — deploying neural networks, configuring LLM serving endpoints and measuring latency under load. This is applied artificial intelligence platform engineering, aligned to Canadian enterprise expectations for measurable outcomes and responsible AI governance.

INF-101

Real-Time Inference Foundations

From C$4,200 · 3 weeks

Establish the core competencies your team needs before touching production endpoints. This module covers model artefact management, REST and gRPC serving patterns, request batching fundamentals and baseline observability for machine learning inference workloads. Participants learn to distinguish sandbox deployments from production-grade model serving configurations and understand how latency optimisation begins at architecture design, not after launch day. Ideal for teams transitioning from experimentation notebooks to scalable AI endpoints. Prerequisites include basic Python fluency and familiarity with at least one ML framework. Delivered hybrid from our Toronto Midtown studio with optional on-site intensives for enterprise cohorts.

INF-201

LLM Serving & Endpoint Design

From C$5,800 · 4 weeks

Large language models demand specialised serving architectures — token streaming, context window management, prompt batching and memory-aware GPU allocation. This module walks through generative AI endpoint design from first principles: choosing between hosted and self-managed LLM serving, implementing streaming responses for conversational AI applications and configuring rate limits that protect infrastructure without throttling legitimate enterprise traffic. Teams leave with a reference architecture for NLP model deployment aligned to responsible AI policies and PIPEDA-conscious data handling. Recommended after INF-101 or equivalent production ML experience.

INF-301

GPU Orchestration & Scaling Lab

From C$6,400 · 4 weeks

GPU resources are the cost centre of modern inference infrastructure. This lab teaches orchestration patterns that match capacity to demand — horizontal scaling for burst generative AI traffic, queue-based scheduling for batch inference and multi-tenant isolation for platform engineering teams serving multiple product lines. Participants configure autoscaling policies, benchmark throughput across model sizes and implement cost-aware routing that keeps p99 latency within SLA without over-provisioning idle clusters. Computer vision and NLP workloads are both addressed with workload-specific tuning exercises.

Workshop-intensive modules

Advanced programmes combine instructor-led workshops with self-paced lab assignments. Our LLM serving workshop (INF-201 companion) gives teams direct experience configuring token streaming endpoints and debugging production inference failures under simulated load — skills that transfer immediately to your AI infrastructure roadmap.

INF-401

Edge AI Deployment Track

From C$5,200 · 3 weeks

Not every inference workload belongs in a centralised cloud cluster. This track covers edge AI deployment — model quantisation for constrained devices, offline-capable inference paths, sync strategies between edge nodes and central observability, and security hardening for distributed endpoints. Manufacturing, logistics and field-service AI automation use cases feature prominently. Teams learn to balance latency gains at the edge against governance complexity and model update propagation challenges inherent to distributed AI infrastructure.

INF-501

Inference Monitoring & Governance

From C$4,900 · 3 weeks

Production ML without governance is a liability. This module establishes inference monitoring dashboards, model drift detection workflows, audit logging for responsible AI compliance and incident response playbooks tailored to Canadian regulatory expectations. Participants implement alerting thresholds tied to business SLAs rather than arbitrary technical metrics, and design access controls that satisfy enterprise security reviews without blocking engineering velocity.

INF-601

Production Inference Certification

From C$9,400 · 6 weeks

The capstone validates end-to-end competency across serve, scale, monitor, optimise, govern and deploy pillars. Candidates architect a complete inference pipeline from model registry through production endpoint, demonstrate GPU orchestration under load test, present observability dashboards to instructors and document responsible AI controls. Certification signals to stakeholders that your team can own production inference operations independently — a credential recognised by our enterprise partner network across the Canadian corporate market.

Scaling module in practice

INF-301 participants work through live scaling scenarios — watching queue depth, GPU utilisation and latency histograms shift in real time as traffic patterns change. This hands-on approach builds the intuition platform architects need when designing AI automation infrastructure that survives production surprises.

Request programme intake