TIER B · ENGINEERING

AI Engineering

Build Production AI Products on Top of Foundation Models

IntermediateCohort-Based15 sessions · 30hDhaka + Global

Build production AI products on top of foundation models — RAG, evals, guardrails.

There is a discipline between 'calls an LLM API' and 'trains models,' and it is where most real AI work now happens: AI engineering — building reliable products on top of foundation models you do not train. It is its own engineering practice, with its own hardest problem (evaluation), its own failure modes (hallucination, prompt injection, cost blowups), and its own infrastructure (vector stores, retrieval, guardrails, LLM observability).

This course teaches that practice end to end. You'll move from raw model APIs to a production-grade AI feature: structured outputs and tool calling, retrieval-augmented generation done properly, rigorous evaluation that catches regressions before users do, guardrails against injection and unsafe output, and the cost/latency engineering that keeps an AI product viable.

Why this course exists

There is a discipline between 'calls an LLM API' and 'trains models,' and it is where most real AI work now happens. Most tutorials stop at a working demo — a chatbot that answers questions from a single document. Production AI requires evaluation, guardrails, cost controls, and observability.

AI engineering is its own practice, with its own hardest problem (evaluation), its own failure modes (hallucination, prompt injection, cost blowups), and its own infrastructure (vector stores, retrieval, guardrails, LLM observability). Without these, every AI feature is a demo that never ships.

This course teaches that practice end to end — from raw model APIs to a production-grade AI feature with eval sets, regression baselines, and cost budgets.

Before

"I called an API and it worked on my laptop. I'm not sure why it fails in production, and the costs are a surprise every month."

After

"I have an eval set and a regression baseline before shipping. Prompts are versioned artifacts. RAG quality is measured. Costs are budgeted by design."

What you'll walk away with

1An eval set and regression baseline in place before any AI feature ships — not "it worked when I tried it"
2Production-grade RAG: hybrid search, reranking, and measured retrieval quality, not naive retrieval that guesses
3Guardrails against prompt injection and unsafe output, plus a cost/latency budget built in by design
4A capstone AI feature shipped with evals, guardrails, cost controls, and observability — defended end to end

Curriculum

15 sessions · 30h across 4pillars · every session ends with a working, reviewable artifact. Expand any pillar for the session lineup.

AI Engineering FoundationsModels, APIs, structured outputs · Week 1–3

▼

Session 1 — What AI Engineering Is (and Isn't)
Session 2 — Foundation Models & Model Selection
Session 3 — The Model API: Chat, Completions & Parameters
Session 4 — Structured Outputs & Tool/Function Calling
Session 5 — Prompt Engineering as Engineering

Retrieval & RAGEmbeddings, vector search, advanced RAG · Week 4–5

▼

Session 6 — Embeddings & Vector Databases
Session 7 — RAG Fundamentals
Session 8 — Advanced RAG (Hybrid Search, Reranking, Query Transformation)

Production ReadinessEvaluation, guardrails, cost, observability · Week 6–9

▼

Session 9 — Evaluation I: The Hardest Problem in AI Engineering
Session 10 — Evaluation II: LLM-as-Judge & Regression Testing
Session 11 — Guardrails, Safety & Prompt Injection
Session 12 — Cost, Caching & Latency Engineering

CapstoneMultimodal and shipping · Week 10–15

▼

Session 13 — AI Observability in Production
Session 14 — Multimodal & Beyond Text
Session 15 — Capstone: Ship an Evaluated, Observable AI Feature

Want the full session-by-session syllabus with objectives and deliverables? Request it here— we'll send it over.

Format & schedule

Delivery	Cohort-based, live, instructor-led
Class length	2 hours per class
Cadence	2 classes/week (4 hours/week)
Total length	15 sessions · 30h over 7.5 weeks
Location	Dhaka + Global (remote seats available)
Prerequisites	Python proficiency, REST API basics. No ML background required.

Pricing

৳4,500

$69 global

Coming soon

This course launches soon. Join the waitlist for early access and launch-day pricing.

FAQ

Do I need a machine learning background?▼

No. This course is for software engineers. You need Python and REST API basics. ML concepts are taught from the ground up where needed.

How is this different from LLM Engineering (Course 5)?▼

AI Engineering treats the model as a black box — you build products on top of APIs. LLM Engineering opens the box — fine-tuning, quantization, self-hosting. Take this one first.

Will I learn to fine-tune models?▼

Not here. Fine-tuning is covered in Course 5 (LLM Engineering). This course focuses on RAG, evals, guardrails, and productionization.

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 →