AI Systems Design: Architecture to Production Scale
From Agent Builder to AI Systems Architect
The capstone — architect production AI platforms at scale.
Completing the Agentic AI cohort leaves engineers able to build, test, and ship a single agent service. This course begins exactly where that one ends. The real architectural challenge is not building one agent — it is designing a system of ten agents, serving thousands of concurrent users, owned by multiple teams, operating under different SLAs, sharing infrastructure, and running against a token budget.
This cohort teaches you to think and communicate at the level of a production AI Systems Architect. Every module addresses a layer of the production stack: service decomposition, inter-service communication, data architecture, Kubernetes operations, SLO design, observability, cost modelling, and security hardening. The capstone is a design exercise — present a complete architecture and defend every decision under challenge.
Completing the Agentic AI cohort leaves engineers able to build, test, and ship a single agent service. This course begins exactly where that one ends.
The real architectural challenge is not building one agent — it is designing a system of ten agents, serving thousands of concurrent users, owned by multiple teams, operating under different SLAs, sharing infrastructure, and running against a token budget. That system does not emerge from adding more LangGraph nodes. It requires a fundamentally different skill: system design.
This course teaches you to think and communicate at the level of a production AI Systems Architect — from a LangGraph node to a Kubernetes HPA policy to an SLO error budget — and justify every decision in writing and in conversation.
Before
"I can build one agent. But scaling to ten agents for thousands of users under SLAs? I'm not sure where to start."
After
"I can design a system of AI services with bounded contexts, Kubernetes autoscaling, SLO-driven operations, and a modelled token budget. I can defend every decision."
- 1Decomposes a platform into bounded, independently deployable AI services — not one agent, but ten under a real budget
- 2Defines SLOs, error budgets, and incident runbooks before the first deployment, not after the first outage
- 3Models token economics and enforces per-tenant spend limits architecturally, not as an afterthought
- 4A defended platform architecture across eight production dimensions — service decomposition, data, Kubernetes, SLOs, observability, cost, security
34h · 9 modules across 4pillars · every session ends with a working, reviewable artifact. Expand any pillar for the session lineup.
Architecture FoundationDecomposition, communication, data · Module 1–3▼
- Module 1 — AI System Decomposition & Bounded Contexts
- Module 2 — Communication Patterns: REST, gRPC & Event-Driven
- Module 3 — Data Architecture for AI Systems
Production OperationsKubernetes, SLOs, observability · Module 4–6▼
- Module 4 — Kubernetes for AI Workloads
- Module 5 — SLO Design for Non-Deterministic Systems
- Module 6 — Observability & Production Monitoring
Cost & SecurityFinOps, threat model, hardening · Module 7–8▼
- Module 7 — Cost Architecture & FinOps for AI
- Module 8 — Security Architecture for AI Systems
CapstoneDesign review and defense · Module 9▼
- Module 9 — Capstone: System Design Review & Defense
Want the full session-by-session syllabus with objectives and deliverables? Request it here— we'll send it over.
| Delivery | Cohort-based, live, instructor-led |
| Class length | 2 hours per class |
| Cadence | Live online, cohort-based — weekly sessions |
| Total length | 34h · 9 modules over 8 weeks |
| Location | Dhaka + Global (remote seats available) |
| Prerequisites | Agentic AI (Course 6) or equivalent hands-on experience with LangGraph, FastAPI, and Docker. |
৳4,500
$69 global
Coming soonThis course launches soon. Join the waitlist for early access and launch-day pricing.
How does this course relate to Agentic AI (Course 6)?▼
Course 6 builds and ships agent systems. This course is the systems-layer counterpart — designing, scaling, observing, and securing them at production load. Take Course 6 first.
Is this a coding course or a design course?▼
Design. The capstone is a structured architecture review, not a code demo. You present a complete system architecture and defend every decision under challenge.
Do I need Kubernetes experience?▼
Helpful but not required. The Kubernetes module teaches what you need from first principles. The focus is on HPA policies and blue-green deployment for AI workloads.
Part of a longer track?
This course is part of the TIER B · ENGINEERING. Explore the full catalog to find related courses that build on each other.
View all courses →