Skip to content
View dunura13's full-sized avatar

Block or report dunura13

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
dunura13/README.md

Hi, i'm Dunura

i'm a 2nd year computer engineering student at the University of Toronto, currently working as a Machine Learning Software Engineer at UTMIST, where i engineer automated pipelines and fine-tune transformer models

iโ€™m also a software developer at utesca, building crowd simulation architectures in c# and unity, and a firmware developer at NeurotechUofT, working on data collection for bci interactions. i love building systems that bridge the gap between hardware, ai, and full-stack web apps.

๐Ÿ› ๏ธ Technical Skills

Languages: Python, Java, C, C++, C#, JavaScript, TypeScript, SQL, HTML/CSS Frameworks: React, Next.js, FastAPI, Flask, PyTorch, YOLOv8, TailwindCSS Developer Tools: Docker, Git, Linux, Azure, Google Cloud Platform (GCP), Unity, AI/ML: LLMs, RAG, Prompt Engineering, Pinecone, OpenCV, LoRA Adapters

๐Ÿš€ Projects

Vibe-Shop | React, OpenAI CLIP, YOLOv8, FastAPI, Pinecone

A multi-modal search interface enabling real-time product discovery via text and image.

  • Containerized Docker backend and an ETL pipeline that improved search relevance by 35%.

Threadlock | Next.js, FastAPI, Gemini API, PostgreSQL, GCP, Jira, Slack API

A GenAI dashboard that visualizes agent activity and automates Jira ticket creation from Slack chats.

  • Implemented RAG workflows to transform unstructured data into structured audit logs.

Self Driving Car | JavaScript, HTML, CSS

A neural network simulation implemented entirely from scratch without external libraries.

  • Built a feedforward neural network to autonomously drive a car through a course, visualizing the network's decision-making in real-time.

Internify | React, Supabase, Lovable, Gemini API

A Gemini wrapper application built to optimize and streamline the internship search process.

๐Ÿ’ผ Experience

Software Developer UTESCA | Toronto, ON | Nov 2025 โ€“ Present

  • Leading the SDLC for a crowd simulation module, translating requirements into scalable C# architecture in Unity[cite: 13].
  • Visualizing simulation outputs in Power BI to provide actionable capacity metrics[cite: 14].

Machine Learning Software Engineer UTMIST | Toronto, ON | Sept 2025 โ€“ Present

  • Fine-tuning PyTorch transformer models using LoRA adapters, boosting F1-scores by 15%.
  • Containerizing Flask/Gradio interfaces with Docker to eliminate environment compatibility issues.

Firmware Developer NeurotechUofT | Toronto, ON | Sept 2025 โ€“ Present

  • Developing Python code to collect and clean IMU/flex sensor data at 200 Hz on Raspberry Pi.
  • Deployed an intent classification model achieving 90% accuracy on raw sensor data.

๐Ÿ“š Education

University of Toronto

  • Bachelor of Applied Science in Computer Engineering, Minor in AI Engineering
  • Sept 2024 โ€“ Apr 2028 | Toronto, ON

๐Ÿ‹ Outside of Code

I lift heavy weights

Pinned Loading

  1. resume-analyzer resume-analyzer Public

    This is a resume-analyzer that utilizes React as the front end, flask for back end, mongoDB to store previous analyzed resumes, and Gemini 2.0 flash lite API to rate provided resumes and give constโ€ฆ

    JavaScript 1

  2. Self-driving-car Self-driving-car Public

    Created a self driving car using a neural network

    JavaScript

  3. Threadlock Threadlock Public

    ThreadLock is an intelligent background agent for Slack that solves "Context Fragmentation."

    Python

  4. Space-Invaders-Remade Space-Invaders-Remade Public

    An advanced version of Space Invaders (1)

    Python

  5. Vibe-Shop Vibe-Shop Public

    Searching Amazon for "mid-century modern chair green velvet" is annoying. Here, users upload a photo of a room they like (from Pinterest/Instagram), and the app finds purchasable items that match tโ€ฆ

    JavaScript 1