Sebastian Czyrny | Computer Engineer

Hi, I'm Sebastian Czyrny.

A

Aspiring computer engineer interested in designing and building large-scale systems, computer networks, and digital systems.

About

I am a graduate student in Electrical & Computer Engineering at the University of Toronto. I enjoy software, computer networks, digital systems development, and CAD for reconfigurable computing. I have experience with front-end web development in AngularJS, HTML, CSS and Javascript; back-end web development in Python, Flask, Django, NodeJS, Java Springboot, and ASP.NET; and databases such as MongoDB and Microsoft SQL Server. Some other technologies I have used are Adobe Coldfusion (a Java Virtual Machine), C/C++, socket programming, Unix development, FPGA development using the hardware description language Verilog, Intel Quartus Prime© Design Software, and embedded programming using the STM32 Nucleo-F446ZE© Microcontroller, Keil uVision5©, and STM32CubeIDE©. I have 2 years of professional experience working around the entire web stack. I am passionate about designing large-scale software systems and implementing them with clean, modular code.

Languages: Python, Java, JavaScript, C/C++, C#, HTML/CSS, Powershell, Coldfusion, Verilog
Databases: Microsoft SQL Server, MongoDB
Frameworks: Flask, Django, NodeJS, TensorFlow, PyTorch, Scikit-Learn, Numpy, Pandas, Bootstrap, Langchain, AngularJS
Tools & Technologies: Git, VSCode, Visual Studio, Eclipse, Microsft SQL Server Management Studio, Keil uVision 5, STM32 Cube IDE
Operating Systems:Windows, Windows Server, Linux Ubuntu & Debian

Computer engineer interested in designing and developing large-scale software systems, computer networks, and digital systems.

Experience

University of Toronto,
Digital Systems, Computer Organization

Teaching Assistant

Facilitated lab sessions for second-year engineering courses on Digital Systems and Computer Organization, providing hands-on guidance and support to students.
Graded assignments, midterms, and lab reports, ensuring timely and constructive feedback to enhance student learning.
Collaborated with fellow teaching assistants and faculty members in regular meetings to discuss student progress and course improvements.
Tools/Software: ModelSim, Intel Quartus Prime, RISC-V, De1-SoC FPGA board

Fall 2024 - Present | Ontario, Canada

MyChart

Student Software Developer

Developed and managed the front-end of a health care information services provider using Javascript, HTML & CSS, and AngularJS
Developed the back-end using Java Springboot framework, ColdFusion Web Development Suite, and connected with a Microsoft SQL Server
Worked alongside a small team of senior software engineers and business anaylsts to design and integrate web pages into the MyChart application
Improved usability through intuitive table filters and pagination, PDF viewing, and data load indicators
Communicated progress on software development through daily meetings and Scrums
Tools: Javascript, HTML & CSS, ColdFusion, AngularJS, Java Springboot, Microsoft SQL Server, Microsoft SQL Server Management Studio

2022 - 2025 | Ontario, Canada

Research

CGRA-ME
University of Toronto

Graduate Research Assistant

Research Focus: Modeling and Exploration of coarse-grained reconfigurable array (CGRA) architectures and algorithms for mapping applications onto them

"RAAP-CGRA: Placement for CGRAs with Restricted Routing Architectures", Co-author

Modeled three increasingly restrictive CGRA architectures within CGRA-ME's C++ framework
Demonstrated that CGRA mapping algorithms that are made aware of the routability of the architecture can significantly improve application mappability for restricted routing architectures over algorithms that are indifferent to it.
Accepted to the CGRA4HPC'25 workshop

"Parallel UGRAMM: Scalable CGRA Mapping", Author

Developed an enhanced CGRA mapper that significantly improves runtime performance without degrading mapping quality.
Explored various parallel programming strategies—both coarse and fine-grained—and demonstrated that a memoization-based approach yielded the most substantial runtime gains. Results showed further speedup with increased CGRA size through deterministic fine-grained parallelization.
Employed Intel VTune to profile performance and identify computational hotspots for targeted optimization.
Achieved a runtime reduction proportional to the size of the dataflow graph (DFG), maintaining full functionality across diverse CGRA architectures.
Under review for ICCAD'25

Fall 2024 - Present

Projects

                
Occasional 
                 A web application that allows event creators to publicize their events to the UofT community. 
                 
Accomplishments
                      
Tools: Python, Python Flask, MongoDB, Bcrypt
Register/login as an event creator or student
Create new events, RSVP, or search.
Collaborated in team of 6 to produce a responsive and useable web app.
Researched UI/UX design principles and incorporated them into a live app.
Administered MongoDB Cloud Database for data persistency
Implemented web Security via the use of Bcrypt for password hashing.

LLM Agent
                 An LLM Agent to Chat with your Shareholder Reports. A Project as part of the UofT Final Year Engineering Capstone course. 
                
Accomplishments
                      
Tools: Python, Langchain, Streamlit, LlamaIndex, SEC EDGAR API
Select Shareholder Reports via PDF upload, fetching from the SEC EDGAR API, and from previously saved.
Choose different prompting strategies and document chunk retrieval mechanisms.
Collaborated in team of 4 to produce an application that utilizes LLMs (both OSS and API Inference) to chat with shareholder reports.
Researched the Natural Language Processing pipeline and incorporated workflow into the application.
Innovated current state-of-the-art prompting strategies with our very own Multi-Stage Prompting technique.
Underwent rigorous Testing of the LLM Agent accuracy through the comparison of human-generated responses with those of the LLM Agent under different prompting and retrieval strategies.
Concluded that no prompting strategy dominates over any other and each has its own use case. The Re-Ranker retrieval strategy performs almost identically to OpenAI's ADA Embeddings Model©.

Live Chat CLI App
                 A chat app that allows user to communicate to each other in real time!
                
Accomplishments
                      
Tools: C, Unix, Sockets, TCP, IPv4
Register/login to the chat app and save credentials to a database.
Create and join sessions hosted by other users on separate devices.
Use TCP and IPv4 to set up connections between client and server.
List all online users and joinable sessions.
Private message other users.

Admin Dashboard Web App
                 A Project Administrator Web Application that allows users to manage their teams, projects, and tickets!
                
                Accomplishments
                Tools: Javascript, HTML & CSS, AngularJS, NodeJS, ExpressJS, MongoDB
Register/login and Javascript Web Tokens for authentication.
Create and manage multiple teams, create projects for each team, and tickets for each project.
Seamlessly switch teams, view statistics, and current team availability
UX/UI design and responsiveness.