Hello

Guneev

Software developer · Ontario, Canada

Scroll

01 / About

Software is the craft of making hard things feel simple.

I'm a computer science student at McMaster who works across cloud and machine learning: designing the infrastructure, building the models, and keeping both fast and reliable in production. Lately that has meant a dual-region serverless platform and a company-wide RAG chatbot at Purolator, a recommender system on Kubernetes, and a home lab running local LLMs.

Studying
Honours Computer Science, minor in Statistics · McMaster, class of 2027
Focus
Cloud infrastructure and ML systems, in equal parts
Based in
Ontario, Canada

02 / Toolkit

The working set.

LanguagesPython

Languages / 5 tools

Pick a layer, then go deeper.

The language behind AI experiments, automation, APIs, and computer vision work.

LanguagesPython · Java · C++ · JavaScript · SQL · Bash
AI / MLPyTorch · Hugging Face · LangChain · RAG · OpenAI API · Amazon Bedrock · Ollama · llama.cpp · MCP
Cloud & DevOpsAWS · GCP · Docker · Kubernetes · Terraform · Linux · Git · GitHub Actions · Prometheus · Grafana
FrameworksReact · Flask · Django · Node.js · Vue.js · FastAPI

03 / Professional experience

Where I build.

PurolatorApplication Engineer Co-opSep 2025 — Apr 2026 · Mississauga, ON
  • Architected a dual-region AWS serverless monitoring platform using CloudFront origin failover, Lambda, and DynamoDB Global Tables, secured with VPC, IAM, WAF, and Okta authentication.
  • Developed Terraform infrastructure and GitHub Actions workflows that automate non-production and production deployments across both regions, with rollback for failed releases.
  • Built a latency monitoring dashboard for 25–30 public, private, and internal applications, used by 900+ people to analyze regional response times and contributing to zero latency SLA breaches over the term.
  • Built a company-wide RAG chatbot with Amazon Bedrock and the Confluence API that answers policy and standards questions for 800+ users.
NovadoxAI Software Developer Co-opMay — Aug 2023 · Toronto, ON
  • Built a GCP-hosted student budgeting web app with a Flask backend and a fine-tuned OpenAI model generating region- and user-specific financial recommendations, with PostgreSQL profile storage for 300+ students.
  • Developed a RAG document question-answering system using LangChain, custom document parsing, OpenAI embeddings, and Vertex AI Vector Search, streaming tokens to bring time-to-first-token under 500 ms.
BlueLogicFull-Stack Software Developer InternJul — Aug 2021 · Dubai, UAE
  • Built the first e-commerce application for a pharmacy brand using React and Flask, with a custom chatbot for buying over-the-counter medication and booking appointments, contributing to a 5% increase in sales.
  • Built a dog breed classifier in Python and PyTorch that reached 86% accuracy on a custom dataset of 500 pictures per breed.
emQubeML Software Developer InternAug — Sep 2020
  • Built Amazon Alexa skills using NLP, ASR, DNNs, and SSML to assist elderly users with daily tasks.

04 / Education + clubs

Learning in public.

Education

McMaster University

Honours Bachelor of Applied Science, Computer Science · Minor in Statistics
Aug 2022 — Apr 2027 · GPA 3.7 / 4.0

Relevant courses

Machine Learning, Natural Language Processing, Algorithms, Operating Systems, Databases, and Computer Networks.

McMaster Computer Science SocietyVice PresidentMar 2024 — present · Hamilton, ON
  • Led a ProjectX 2024 ML research competition team on improving co-learning in multimodal AI and co-authored the paper “MoonWhisper: Using Audio Inputs to Augment Vision-Language Models”.
  • Mentored 8 undergraduate students and organized technical events and resources for the computer science community.
McMaster Solar CarSoftware LeadMar 2024 — present · Hamilton, ON
  • Built a vehicle performance and strategy optimization system that models conditions and optimizes target speeds, improving simulated performance by 15%, with a GUI for input configuration and simulation visualization.
  • Designed, developed, and hosted the team's public website using Vue.js and JavaScript.
McMaster Cyber Security ClubOutreach ExecutiveSep 2024 — present · Hamilton, ON
  • Managed university, sponsor, and event communications and coordinated cybersecurity events.
  • Competed in 3 CTFs with a top-3 CyberSci finish.

05 / Selected projects

Made with intent.

Things I built to understand how they work, then kept running.

01 / Infrastructure · Local inference

Home Lab & Self-Hosted AI

A private cloud across three servers running five containerized workloads with automated updates, hourly health checks, and self-healing recovery. Two local LLMs on Ollama and llama.cpp drive six MCP tools that automate my daily workflows, all reachable securely over TLS and a VPN.

3
Servers
5
Workloads
64
Tokens / sec
  • Linux
  • Docker
  • Ollama
  • llama.cpp
  • MCP
  • Prometheus
  • Grafana

02 / Recommender systems · RAG

Movie Recommendation & RAG Platform

A sequential two-tower recommender with a causal Transformer user tower, trained on 15.5M interactions and ranked against a 54K-movie catalog. A RAG pipeline explains every suggestion. It first ran on Kubernetes with Terraform-managed AWS infrastructure; the cluster is now switched off, and the live demo runs the model right in your browser.

0.135
NDCG@10
5.8×
vs. baseline
<120ms
p95 latency
  • Python
  • PyTorch
  • FastAPI
  • Kubernetes
  • AWS
  • Terraform

03 / Computer vision · Paper build

YOLOv3 from scratch

The YOLOv3 paper implemented in PyTorch: Darknet-53, the three-scale detection head, the training loss, multi-scale training, and COCO evaluation. The authors' released weights load unchanged and reproduce the reference detections.

  • Python
  • PyTorch
  • Darknet-53
  • COCO

04 / Research paper · Multimodal AI

MoonWhisper

Co-authored ProjectX 2024 research that teaches the Moondream vision-language model to hear by feeding Whisper audio embeddings into its language model. It reached 48.16% accuracy on OmniInstruct and held steady when the audio was replaced with noise.

  • PyTorch
  • Whisper
  • Moondream
  • Phi-2

05 / Hackathon · Security

NδtHacked

A personal internet security suite built at DeltaHacks XI: a curated privacy report, password generator and encrypted vault, suspicious-link checker, and VPN detector, paired with Arduino hardware.

  • Python
  • Cohere API
  • Arduino

06 / NLP · Competition design

Reddit personal-safety classifier

An NLP task and Codabench competition that rates how threatened someone's life is from first-person Reddit posts on a five-level scale, including the scraper, labelled dataset, and scoring pipeline.

  • Python
  • BERT
  • Reddit API
  • Codabench

06 / Contact

Have a good idea?
Let's make it real.

guneevwork@gmail.com