AI ENGINEER

Explore the AI engineering landscape and define how machine learning translates real business problems into production systems. Learn to distinguish core industry roles, evaluate when AI is appropriate, and frame a capstone problem using a structured ML approach.

• Role definitions
• Go/no-go framework
• Problem mapping
• Capstone kickoff

Learn how to build reproducible ML development environments using industry-standard tools. You’ll set up containerized workflows, implement experiment tracking, and structure projects for scalable development.

• Docker basics for local development
• Experiment tracking
• Project structure conventions

Discover how to explore, clean, and structure raw data into reliable inputs for machine learning. You’ll build automated workflows that make datasets usable, traceable, and production-ready.

• Data exploration
• Data cleaning
• Data auditability
• ETL pipeline

Learn how to evaluate dataset integrity and prepare robust data splits for training and testing. You’ll apply statistical checks to ensure your data is reliable before modeling.

• Data validation/testing
• Historical distribution checks
• Identifying and measuring "good" vs. "bad" data
• Train/val/test split strategies

Assignment #1:
Create and evaluate a task dataset.

Explore how to build strong baseline models that set performance benchmarks before optimization. You’ll construct structured ML pipelines and apply foundational feature engineering techniques.

• Baseline importance
• Traditional ML models
• Feature engineering basics
• ML pipelines

Learn how to evaluate model performance using the right metrics, validation strategies, and tradeoff analysis. You’ll interpret results and decide what “good” actually looks like for your problem.

• Classification metrics
• Metric tradeoffs
• Primary vs. secondary metrics
• Evaluation methodology

Assignment #2:
Create and evaluate a baseline model.

Discover how to systematically find where your model breaks and why. You’ll use structured error analysis and data slicing to uncover failure patterns and improve performance.

• Error-analysis loop
• Data slicing to find failure modes
• Strategies by model type

Learn how to improve models through structured iteration while avoiding overfitting and wasted compute. You’ll refine models using principled tuning and clear stopping signals.

• Macro training loop
• Hyperparameter tuning
• Overfitting detection and mitigation
• Stopping criteria

Assignment #3:
Analyze baseline model errors and iterate.

Explore how large language models work and how to reliably shape their outputs. You’ll design prompts, manage cost constraints, and structure model responses for real applications.

• Core concepts
• Prompt patterns
• Cost modeling

Learn when to fine-tune models versus using prompting or RAG. You’ll evaluate LLM performance using modern assessment methods and optimization techniques.

• When to fine-tune vs. prompt/RAG
• Deep learning basics for fine-tuning
• Fine-tuning evaluation
• Parameter-efficient methods

Assignment #4:
Develop and iterate on prompt for LLM-based model.

Discover how to extend model knowledge using retrieval-augmented generation. You’ll build vector-based retrieval systems and implement memory-aware LLM applications.

• RAG architecture
• Vector databases
• Chunking strategies, reranking
• Evaluation of RAG
• Memory as an application of RAG

Learn how to build multi-step LLM applications using tools, agents, and safety guardrails. You’ll design systems that manage context and integrate external capabilities.

• Agentic architecture
• Tool integration
• Guardrails
• Context management

Assignment #5:
Build and iterate on a RAG subsystem.

Explore how ML systems are containerized and prepared for scalable deployment. You’ll optimize Docker workflows and learn core Kubernetes concepts.

• Production Dockerfiles
• Container orchestration
• Kubernetes concepts intro

Learn how to take models from training to production using CI/CD and deployment strategies. You’ll connect model versioning, testing, and deployment pipelines.

• CI/CD pipelines
• Artifact and model registries
• Production readiness
• Deployment strategies

Assignment #6:
Deploy a model with a simple UI via Docker.

Discover how to detect drift and monitor model performance in production. You’ll build feedback loops that keep systems reliable over time.

• Data drift vs. concept drift
• Drift detection
• Alerting pipelines, retraining triggers, continuous evaluation
• Feedback loops
• Observability

Learn how to reduce inference cost and improve system performance. You’ll apply compression and infrastructure strategies to make models efficient at scale.

• Model compression
• Latency budgeting, caching strategies, request batching
• Infrastructure optimization
• Cheap filters

Assignment #7:
Add monitoring to a Docker deployment and interpret the results.

Explore how to evaluate fairness, privacy, and compliance in ML systems. You’ll implement safeguards and document model behavior responsibly.

• Bias detection & fairness metrics
• Privacy engineering
• Pre-ship validation
• Model documentation

Learn how to translate model performance into business impact. You’ll connect technical outputs to ROI and stakeholder decision-making.

• ROI framework
• Build vs. buy decision framework
• Translating technical results for stakeholders
• Mapping engineering results

Integrate full ML systems and test them through peer review and adversarial evaluation. You’ll stress-test your application for real-world conditions.

• End-to-end integration
• Peer red-teaming
• Error handling, graceful degradation

Assignment #8:
Finalize project implementation in preparation for the presentation.

Present your complete AI system through a live demo. You’ll showcase functionality, technical design, and measurable business impact.

• Live presentations