Course Overview

 Foundation of AI-Native Product Engineering 

Instructor(s):

• Dr. Md Alamgir Kabir, Assistant Professor & Coordinator, MIS. Dept. of CSE Daffodil International University
• Ovi Shekh, AI Native Engineer Founder Of Arklab AI

Duration: 48 Hours

Mode of Delivery: Blended Learning (Technical Laboratory Coding + Continuous Hands-On Applied Architecture Sprints)

Course Description

This foundational course is designed to transform traditional deterministic software developers into foundational AI engineers. It provides hands-on fluency in prompt orchestration pipelines, context-window token management, and localized data retrieval architectures. Learners will move beyond basic AI chat interfaces to programmatically engineer stable, predictable software systems that interface securely with Large Language Models.

Evidence of Demand

Rapid digitization across enterprise engineering platforms in Bangladesh has sparked an intense, immediate need for software developers who can go beyond using basic AI chat interfaces and instead programmatically engineer stable, predictable software systems that interface securely with Large Language Models.

Purpose and Objectives

Purpose: To transform traditional deterministic software developers into foundational AI engineers by providing hands-on fluency in prompt orchestration pipelines, context-window token management, and localized data retrieval architectures.

Objectives:

• Master the physical configuration of prompt templates, structured output constraints, and native function calling inputs.
• Build from scratch a live document parsing, chunking, and metadata injection pipeline.
• Implement a working Retrieval-Augmented Generation (RAG) system utilizing open-source local vector storage.

Course Content & Class Plan (Modules)

• Module 1: AI-First Thinking & Context Mechanics: Hands-on evaluation of context window limits, token calculation debugging, and tracking API runtime latencies.
• Module 2: AI-Native Product Idea Generation: Evaluating deterministic software boundaries vs. probabilistic AI-first product opportunities; mapping real-world problems to functional AI ideas.
• Module 3: Prompt Engineering with Tools and Functions: Writing, testing, and versioning system prompts; configuring explicit function-calling architectures for predictable JSON outputs.
• Module 4: Scraping & Parsing (AI Ready Data Platform): Writing data ingestion scripts to scrape text and parse unstructured files (PDFs, Markdown); implementing semantic document chunking.
• Module 5: RAG Models (AI Ready Data Platform): Initializing/managing open-source local vector databases (ChromaDB); generating text embeddings and implementing vector similarity search.
• Module 6: Visualization and Presentation: Designing clean user interfaces to expose semantic data metrics, retrieval histories, and model outputs.
• Module 7: Capstone Project - Idea to Pitch Deck: Consolidating validated product concepts into technical pitch decks and presenting architectural blueprints.
• Module 8: Building a PRD and Build Prompt: Writing a technical Product Requirement Document (PRD) mapped to AI system behaviors and translating it into an immutable master prompt stack.
• Module 9: Using AI Native Tools for Building Scalable Applications: Orchestrating components into an end-to-end web application using rapid prototyping libraries (Streamlit or Gradio) and deploying a live sandbox.

Practical & Field Work

• Technical Laboratory Coding: Continuous hands-on applied architecture sprints.
• API & System Testing: Profiling computational overhead, testing system response degradation under max token pressure, and managing developer keys.
• App Deployment: Orchestrating ingestion, prompt templates, and generation components into an end-to-end web application deployed on a local network environment.
• Final Code Defense: Timed practical presentation of a working Streamlit-backed AI app judged on database consistency, clean execution, and script structure.

Learning Outcomes

Upon successful completion, learners will be able to:

1. Enforce strict JSON object structures on language model inference loops to guarantee software API stability.
2. Develop and deploy a working data parsing, embedding generation, and metadata indexing code routine.
3. Initialize, build, query, and maintain localized vector database arrays (ChromaDB).
4. Package and run an interactive, data-grounded AI-native application via an accessible web-based graphical interface.

Target Audience & Requirements

Target Audience:

• Software engineers, full-stack web and mobile developers, database administrators.
• Senior undergraduate university students (CSE, SWE, CIS) aiming to specialize in AI engineering pipelines.

Entry Requirements:

• Foundational proficiency in Python programming syntax (loops, dictionaries, functions, and standard object-oriented concepts).
• Core understanding of client-server operations and basic web APIs.
• Minimum Age: 18 Years.

Career Pathways

• Earns professional micro-credential credits tracking toward advanced systems engineering.
• Successful completion of this specific foundation block serves as the mandatory technical prerequisite to unlock entry into the "Advanced of AI-Native Product Engineering" course.

Assessment Criteria

Aligned fully with Outcome-Based Education (OBE) weightage models:

• Continuous Laboratory Code Project Commits (Git Lab Tracking): 40%.
• Final Functional Core Product Deployment & Code Defense: 40% (Timed practical presentation of a working Streamlit-backed AI app).
• Functional Data Ingestion and Retrieval Pipeline Test: 20%.

Tools & Resources

• Facilities & Equipment: Multimedia computer laboratory setups, persistent internet routing.
• Software & Integrations: Local development IDE installations (VS Code/Jupyter), API environment tokens, integration with the DIU Learning Management System (LMS).
• Class Size: Maximum 25-30 participants per operational batch section.

Financial Information, Certification

• Tentative Course Fee: 3,000 BDT
• Certification: Earns professional micro-credential credits upon completion.