OLEKSANDR
NEKRASHEVYCH

SOFTWARE ENGINEERING • ROBOTICS • AUTOMATION

+27 66 558 3327 shonnekrashevych@gmail.com

[ CLICK TO ENTER ]

ABOUT

I am a Computer Engineering student focused on building systems across robotics, embedded computing, and intelligent software.


My work spans robotics, computer vision, embedded systems, and backend engineering, with a strong emphasis on building practical, end-to-end systems that integrate hardware and software.

PROJECTS

VIEW PROJECT
Facial Feature Extraction & Robotic Portrait Sketching System
Computer vision and embedded robotics system capable of extracting facial features from images and reconstructing portraits using a custom-built 3-axis drawing machine with motorized rail control and coordinate-based motion planning.
Python C Computer Vision Image Processing Embedded Systems Robotics Motion Control
VIEW PROJECT
Autonomous Multi-Colour Line-Following Robot
Autonomous embedded robotics platform capable of dynamically tracking red, green, blue, or black race lines through selectable real-time colour sensing, custom photodiode circuitry, and low-level control firmware. Semi-finalist in the 2024 TUKS Robot Race Day competition.
Assembly Signal Processing Control Systems Embedded Systems Robotics
VIEW PROJECT
Vision-Based Autonomous Chess Agent
Autonomous chess-playing system that operates directly on a graphical chess interface using real-time computer vision for board state reconstruction, a Stockfish-based engine for move generation, and OS-level input automation for execution. The system performs full visual state extraction from screen data and converts it into structured FEN representations without using APIs or web scraping.
Python PyTorch OpenCV Computer Vision State Reconstruction Machine Learning System Automation
VIEW PROJECT
Conversational Workflow Agent (WhatsApp LLM System)
Built an LLM-powered conversational agent that operates over WhatsApp to handle customer interactions, classify intent, and route requests to appropriate internal workflows. The system engages users in natural conversation, extracts requirements, and generates structured summaries of conversations for automated routing to relevant business processes such as support, complaints, or service requests.
Node.js LLM Integration Messaging Systems Workflow Automation Backend Systems
VIEW PROJECT
Autonomous Context-Aware Agent System (Ongoing Independent Research Project)
Ongoing independent research project focused on the development of a modular autonomous agent system with persistent memory, contextual awareness, and adaptive behavioural patterns driven by accumulated experience and environmental context. The project is being actively built to explore how a continuous agent can perceive context, retain long-term memory, and adapt behavioural tendencies over time through memory-driven and state-based reasoning mechanisms.
Python LLM Systems Agent Architectures Memory Systems Async Runtime Autonomous Agents

TECHNICAL SKILLS

Languages: Python, C, C++, Java, JavaScript, Lua, Assembly
Technologies: Node.js, Docker, MySQL, PostgreSQL, MongoDB, Git, OpenCV, MATLAB, LTspice, HTML, CSS, Webflow, Roblox Studio, Godot, Embedded Systems
Concepts: Software Architecture, Computer Vision, Robotics, Autonomous Systems, Control Systems, State Estimation (KF/EKF), Real-Time Systems, Signal Processing, Embedded Systems, Game Development

EXPERIENCE

Cheery Robot Academy
Programming & Robotics Instructor • April 2024 – Present
  • Instruct programming and robotics across middle school, high school, and university-level students, covering foundational to intermediate concepts in software engineering and embedded systems.
  • Teach core programming principles using Java, Python, and C++ for academic development, alongside applied development in game development environments such as Roblox Studio and Godot.
  • Deliver hands-on robotics and electronics instruction focused on microcontroller-based systems, sensors, and introductory embedded systems design.
  • Guide students through practical software projects spanning game development, automation, and interactive system design, with emphasis on real-world application of programming concepts.
Upwork
Freelance Software Developer • July 2024 – July 2025
  • Designed and developed 80+ SQL exercises for a PostgreSQL-focused Udemy course used by over 5,000 learners, covering core database concepts including joins, indexing, constraints, subqueries, and query optimization.
  • Contributed to curriculum structure and learning progression design, ensuring material followed a coherent difficulty gradient and aligned with intended educational outcomes.
  • Acted as a technical bridge between subject matter expertise and learner understanding by identifying implicit assumptions in course content and adapting explanations to better match beginner and intermediate student perspectives.
  • Collaborated with the primary course creator to refine instructional clarity, improve accessibility of technical concepts, and ensure consistent understanding across diverse learner backgrounds.
  • Delivered production-ready educational content as part of a freelance engagement, resulting in a 5-star client rating on Upwork.
RunJumpyFly
Frontend & Embedded Systems Intern • July 2024 - August 2024
  • Implemented a responsive single webpage for a financial consultancy website from Figma designs using Webflow, HTML, CSS, and JavaScript, translating UI specifications into a functional frontend implementation.
  • Developed reusable frontend components and interactive UI elements to improve layout consistency, navigation structure, and cross-device responsiveness within a structured design-to-code workflow.
  • Implemented embedded control logic for an ESP32-based LED strip system, designing programmable lighting sequences as part of exploratory embedded systems development work.
KeyStone Electronic Solutions
Backend Vacation Work Intern • July 2022
  • Worked on a legacy Java Spring Boot application, resolving dependency and library version incompatibilities to restore functionality on modern Ubuntu Linux environments.
  • Assisted in integrating a MySQL database for persistent data storage and retrieval within the existing backend system.
  • Containerised the application using Docker and aligned compatible dependency versions to ensure stable and reproducible deployment across Linux systems.

EDUCATION

University of Pretoria
BEng Computer Engineering • Expected December 2026
Relevant Coursework: Embedded Systems, Robotics, Control Systems, Digital Signal Processing, Sensor Fusion, State Estimation (KF, EKF, UKF), Computer Vision, Machine Learning & Neural Networks, Algorithms, Electronics, Linear Algebra, Probability & Statistics

CERTIFICATIONS

Cisco Networking Academy
December 2020 – January 2024
  • CCNA: Introduction to Networks
  • CCNA: Routing and Switching Essentials
  • CCNA: Switching, Routing, and Wireless Essentials
  • CCNA: Enterprise Networking, Security and Automation
  • CCNP Enterprise: Core Networking

Conversational Workflow Agent for Customer Service (WhatsApp LLM System)

Developed a conversational workflow agent that operates within WhatsApp to handle customer interactions, route requests, and manage escalation flows in real time. The system is designed as a workflow orchestration engine rather than a traditional chatbot, focusing on structured decision-making and automated routing logic.

The agent processes incoming messages, performs intent classification, and dynamically routes interactions into multiple operational flows including automated support responses, escalation to human operators, and structured project request generation. Project-related inputs are converted into formatted summaries and forwarded to internal teams for review and action.

  • Intent classification and conversational routing logic implemented in a Node.js backend
  • Automated response generation for customer support queries
  • Human escalation pipeline with full conversation context handoff
  • Structured project request extraction with formatted internal summaries
  • WhatsApp messaging integration via external API layer

This system demonstrates backend workflow engineering, LLM-assisted conversational routing, and real-world automation of customer communication and internal decision pipelines.

VIEW GITHUB

Facial Feature Extraction & Robotic Portrait Sketching System

Designed and developed a complete autonomous portrait reconstruction system capable of extracting facial structures from images and physically reproducing them using a custom-built 3-axis robotic drawing platform.

The project combined first-principles computer vision, embedded systems, coordinate transformations, motion control, and path planning into a unified end-to-end pipeline capable of converting facial imagery into executable robotic drawing instructions.

  • Implemented custom facial feature extraction and image processing algorithms from first principles
  • Designed and constructed a motorized 3-axis rail-based drawing system for coordinate-driven sketch reproduction
  • Developed embedded motion-control firmware for synchronized robotic positioning and drawing execution
  • Built a real-time coordinate transformation and path generation pipeline for image-to-motion conversion
  • Integrated onboard processing, robotic control, and autonomous sketch reconstruction into a unified system
GITHUB UNAVAILABLE Repository unavailable due to University of Pretoria project ownership and IP restrictions

Vision-Based Autonomous Chess Agent

Built a fully autonomous chess agent that operates on a live graphical chess interface using a closed perception–decision–action pipeline.

The system performs screen-based board detection and reconstructs game state through coordinate mapping and FEN conversion, which is then passed to a Stockfish engine for optimal move generation.

  • Screen-based chessboard detection and structured board state reconstruction from visual input
  • FEN generation and internal board representation for engine compatibility
  • Stockfish integration for move planning and decision-making
  • CNN-based piece classification to improve robustness of visual state extraction
  • Grid-to-screen coordinate mapping for precise move execution
  • OS-level input automation enabling end-to-end autonomous gameplay loop
VIEW GITHUB

Autonomous Personality-Driven Agent System (Ongoing Independent Research Project)

Ongoing independent research and development project focused on building a modular autonomous agent system with persistent memory, contextual awareness, and adaptive behavioural state management. The work is aimed at exploring long-term agent identity, continuous environmental awareness, and behaviour adaptation over time rather than a static application.

The system is being actively developed to understand how an agent can maintain and evolve internal state over time through memory, contextual input, and environmental signals, enabling behaviour that adapts through accumulated experience and state-driven reasoning. Current work focuses on memory persistence, contextual reasoning, and early-stage perception and interaction capabilities.

  • Persistent memory system for long-term state retention and retrieval (in development)
  • Context-aware reasoning layer for interpreting environmental and conversational context (in progress)
  • Screen-based perception pipeline using image processing techniques (early-stage implementation)
  • Experimental autonomous action execution framework for system interaction (prototype stage)
  • Async modular agent architecture for concurrent cognitive processes (under development)

Autonomous Multi-Colour Line-Following Robot

Designed and implemented an autonomous embedded robotics system for the TUKS Robot Race Day competition, achieving a semi-finalist placement through robust real-time navigation and adaptive line tracking.

The system was developed entirely from first principles with no external libraries, including custom photodiode-based sensing circuitry, signal interpretation, and low-level assembly firmware for deterministic motor control and sensor processing.

  • First-principles photodiode-based line detection and analog signal processing (no external libraries)
  • Assembly-level embedded firmware on PIC microcontroller for real-time control
  • Multi-colour track selection and adaptive navigation logic (red, green, blue, black)
  • Closed-loop motor control with low-latency response system
  • Competition-validated autonomous navigation system (semi-finalist, TUKS Robot Race Day)
GITHUB UNAVAILABLE Repository unavailable due to University of Pretoria project ownership and IP restrictions