MSc-FinalExam

The Final Chapter of Story

Advanced Software Technology

Software development workflows and methodologies

cowboy coding

waterfall

agile

TDD

scrum

kanban

User eXperience

what it is

its main parts

Requirements: use case diagrams

user stories

BDD

DevOps

version control systems

git

CI/CD

automatic testing

the three levels of MLOps

OOP basics

classes and objects

class diagram

fields

methods

visibility

encapsulation

relationships between classes

inheritance

polymorphism

SOLID principles

OOP design patterns

Strategy

Iterator

Factory

Singleton

Observer

Decorator

Composite

Embodied Intelligence

Embedded Systems

Main building blocks and their relationship with the environment

Control loops (open/closed)

Signal (Types of signals (based on the time/value quantization)

Signal preprocessing methods; Filter types)

Microprocessor as the computational unit of the Embedded Systems

Most used peripherals; I/O - purpose and usage

Timer - purpose and usage

AD/DA converters – purpose and usage

Comparators

Communication protocols

Real timeness

Sensors

Mostly measured physical quantities

Measurement in mechanics

Indirect measurement idea

Principle of the optical encoders

Actuators

Main energy sources by type used

Power machines vs actuators

Positioning

DC motor

BLDC motor

Stepper motor

Linear motor

Special actuators (piezo motor, memory alloy, MEMS)

Principle of the servo motor

Ethorobotics

Social robots

Industrial robots

Uncanny valley

Main fields of application of social robotics

Communication modalities in interactions

Attachment and the Ainsworth Strange Situation Test

Cognitive Robotics

cognitive architectures

adaptivity

Braitenberg vehicles

cognitive model of iPhonoid

robot pianist

Evolutionary Robotics

robot path planning

workspace optimization

estimation of kinematic chain

welding robot

Biologically-inspired robot locomotion

evolutionary-based locomotion

neurooscillator-based locomotion generation

evolving a sensory-motor interconnection structure

Natural Language Processing & Foundation Models

Tokenization

Whitespace/whole-word tokenization

regular expressions

edit distance

subword tokenization

Byte Pair Encoding

Language modeling

The goal of a language model

continuation probabilities

start and end tokens

Markov assumptions in N-gram models

LM evaluation

Transformer-based Language Models

The bottleneck problem of RNNs

Multi-head attention

Transformers

masking

positional encoding

teacher forcing

self-attention

cross-attention

properties of the GPT and BERT model families

Modern LLM training methods

Alignment

instruction and chat models

steps and data types of training

quantization

distillation

adapters and prompt tuning

LLM Applications

base principles of prompt and answer engineering

tooling

embedding models

retrieval augmented generation

vector-similarity search

ReAct-style agentic systems

Multimodal Language Models

InfoNCE

CLIP

ViT

tokenization with VQ-VAE and dVAE

encoder-decoder and prefix decoder visual-language models

Text-to-image methods

Autoregressive text-to-image models [DALL-E]

the base principle of diffusion models [graphical models only, no formulas]

Latent Diffusion Models

Classifier-free guidance

Speech to text processing

STT task definition

speech signal processing [sampling, Fourier transform, Mel spectrum]

acoustic modeling

combining language models with acoustic models

connectionist temporal classification