MM●
Calgary, Canada

Mehdi
Marzban.

Mehdi in graduation attire holding his degree and flowers
University of Calgary · graduation

Software engineering.
Machine learning research.

I'm a University of Calgary graduate with a thesis-based M.Sc. in Electrical Engineering and a B.Sc. in Software Engineering, with a minor in Biomedical Engineering.

Research ↘Projects ↗
M.Sc. Electrical EngineeringB.Sc. Software EngineeringBased in Calgary Open to remote opportunities
Open to relocation
1-2B.Sc.

University of Calgary · 2018–2022

Software Engineering

Minor in Biomedical Engineering

2-1

2026–present

Outlier · Contractor

AI Response Evaluator

Contractor engagement, not a full-time or part-time position.

  • Identify difficult cases and edge cases that may cause models to produce incorrect responses.
  • Formulate evaluation criteria and judge model responses against the task requirements.
2-2

May 2022–December 2025

University of Calgary · Calgary, Alberta

Graduate Research Assistant

  • Worked across the PRIME and SHAPE projects and MATLAB-to-Python conversion of the RTTM application throughout the research assistant period.
  • Built Python/Flask backend services for pipeline simulation and service integration for at least one year of the funded research period.
  • Analyzed CSV and Excel datasets with Python and Matplotlib to produce plots and reports for engineering research.
  • Collaborated through private GitLab repositories, project-ticket reviews, and weekly Scrum meetings to coordinate software implementation with the research team.
View research projects ↘
2-3

May 2022–April 2024

University of Calgary · Calgary, Alberta

Graduate Teaching Assistant

  • Supported one engineering course per semester, helping students work through laboratory exercises and assignments.
  • Graded assignments and examinations to assess students’ understanding of course material.
  • Mentored capstone students through milestone feedback, technical advice, and grading submissions.
Earlier experience
2-4

January 2018–January 2020

Real Canadian Superstore · Part-time

Cashier

Customer service and checkout work in Calgary.

3-1 / M.Sc. thesis / University of Calgary · 2026

Enhancing Leak Detection in Oil and Gas Pipelines with Multi-Task LSTM Model

My master's research studies how a model can detect a pipeline leak and estimate its location from multivariate time-series data.

The work includes preprocessing, model development, and evaluation using simulated pipeline data, with a comparison between multi-task LSTM and 1D CNN models.

PythonTensorFlow / KerasTime series
Read the published thesis ↗
4-22025

Journal article

Volumetric Changes in Cerebellar Transverse Zones: Age and Sex Effects in Health and Neurological Disorders

Farshid Ghiyamihoor, Payam Peymani, Jarrad Perron, Azam Asemi-Rad, Mehdi Marzban, et al.

Human Brain Mapping · 46(6), e70214

DOI: 10.1002/hbm.70214
Full author list

Farshid Ghiyamihoor, Payam Peymani, Jarrad Perron, Azam Asemi-Rad, Mehdi Marzban, Aashka Mohite, Karen Ardila, Bara Aljada, Asghar Marzban, Mehnosh Toback, Sherif Eltonsy, Ji Hyun Ko, Tabrez J. Siddiqui, Christopher J. Steele, Jiming Kong, Mario Manto, M. Ethan MacDonald, Jason S. Gill, Roy V. Sillitoe, Fuat Balcı, Iman Beheshti, and Hassan Marzban

PubMed record ↗
5-1

Pipeline visualization

PRIME

Explore a pipeline route in ArcGIS and Cesium. Select pipeline nodes and review leak, stress, risk, and what-if interfaces.

Front-end reconstruction of a collaborative master's project at the University of Calgary's Pipeline Engineering Centre (PEC).

Human-X Interaction Lab (HXIL)PEC — Pipeline Engineering Centre

Project funding: Alberta Innovates ↗

JavaScriptArcGISCesium
Open demo ↗
Project details +

The demo renders the source route with selectable nodes and seven station-to-station spans. Engineering screens reproduce the original interface; the original simulation services are not connected.

Thesis section 5.1 ↗
My contribution
  • Built the original PRIME web application, including its React frontend and Flask backend.
  • Set up communication servers and HTTP integration with the existing real-time transient model (RTTM) engine.

The linked demo is a frontend reconstruction of that research application. The RTTM engine was developed by others.

PRIME web app showing the pipeline route on an ArcGIS mapOpen PRIME
5-2

XR digital twin · University of Calgary

SHAPE

Safe Hydrogen Agile Pipeline Engineering: an immersive digital twin of the University of Calgary's APRIL facility for pipeline visualization, leak simulation, and operator training.

My contributions included Python/MATLAB backend components and REST API integration.

University of Calgary · Pipeline Engineering Centre (PEC), HXIL, and Serious XR (SEER) Lab.

Presented at IEEE ISMAR 2025 in Daejeon, South Korea.

Human-X Interaction Lab (HXIL)PEC — Pipeline Engineering CentreSEER — Serious XR Lab

Project funding: Alberta Innovates ↗

Python / MATLABREST APIsUnity / XR
Thesis section 5.2 ↗
Project details +

The research prototype combines an APRIL facility scan, a CAD pipeline model, sensor visualization, and an embodied virtual agent. The conference presentation describes sensor-to-database and simulation workflows, the Unity XR interface, and a planned usability study.

Developed at the University of Calgary's Pipeline Engineering Centre (PEC), in collaboration with the Human-X Interaction Lab (HXIL) and the Serious XR (SEER) Lab. The demo documents the prototype; full integration of the thesis ML models is discussed as future work.

My contribution
  • Set up backend services and communications for the collaborative XR pipeline digital twin.
  • Worked on Python/MATLAB backend components and REST API integration.
  • Presented the co-authored workshop paper at IEEE ISMAR 2025 in Daejeon, South Korea.

My Unity involvement was mainly testing; my primary development role was on the backend.

SHAPE prototype demo 54 seconds · no audio

A walkthrough of the XR pipeline model, sensor overlays, and flow visualization.

Presentation visuals +

An Immersive Digital Twin with Virtual Agent Interface for Pipeline Leak Simulation and Monitoring. Conference presentation, IEEE ISMAR 2025.

5-3

Agentic document review

Document Review Agent

Compare documents, investigate missing evidence, and inspect the sources behind each finding. A LangGraph workflow lets Gemini choose when to search, read a page, or finish a review.

An evolution of Document Evidence Assistant · working preview.

PythonLangGraphGeminiStreamlit
Open demo ↗
Project details +

Supports PDF, TXT, and Markdown documents with source and page references, citation validation, bounded model requests, and a visible activity trace. The hosted demo uses Gemini; the separate Windows and Mac desktop edition runs Ollama with Qwen2.5 7B locally. Check cited passages before relying on findings.

View source code ↗
My contribution

Directed requirements and reviewed an AI-assisted document-review workflow. I tested document selection, answer coverage, citations, and unsupported claims, and used those findings to request changes to the application.

Illustrated current Document Review Agent interface showing local-edition downloads and the Gemini review flowOpen Document Review Agent
5-4

Document retrieval

Document Evidence Assistant

Upload a PDF, ask a question, and inspect the supporting passages. Optional Gemini answers include references to the source document.

PythonStreamlitGemini
Open demo ↗
Project details +

Retrieval and citation validation keep supporting text visible. Development cases cover table values, missing information, and conflicting statements. Generated claims still require human review.

My contribution

Tested the assistant against my thesis using questions about table values, missing information, and conflicting statements. I reviewed generated answers and source passages and used the results to guide improvements. Development was AI-assisted.

Document Evidence Assistant web app with PDF upload and cited-answer controlsOpen Document Evidence Assistant
5-5

Android · undergraduate capstone

HandiApp

A mobile app for finding and sharing accessibility information about places. Users can search a map, read community reviews, and describe their own experience of an establishment.

Collaborative software engineering capstone · 2022

Kotlin / AndroidAWS AmplifyGoogle Maps
Project details +

The Android app supports place search, adding and managing accessibility reviews, keyword filtering, and text-to-speech. The team used Kotlin and XML for the interface, Google Maps and Places for location search, and AWS Amplify, AppSync, DynamoDB, and Cognito for data and authentication.

These screenshots document the capstone application. A current public app release is not linked.

My contribution

Worked on the frontend, backend, and database for HandiApp with my capstone team. We shared implementation across the application, including the Kotlin/XML interface and the AWS-backed data and review workflows.

5-6

Spanish learning · working prototype

SoloPoco

A small Spanish learning app for absolute beginners. Listen to short conversations, practise useful phrases, recall them from memory, and revisit them with spaced review.

AI-assisted development · public demo, no account required.

React / TypeScriptCloudflare Workers / D1Gemini
Try SoloPoco ↗
Project details +

Six lessons combine saved Spanish audio, slow playback, transcripts, comprehension checks, and written recall. A server-side Gemini tutor provides short replies, translations, and corrections within request limits. New tutor audio is generated on the visitor's device after an automatic voice download.

Cloudflare D1 stores progress using a signed guest cookie. Progress stays with that browser; clearing cookies starts a new guest. This is a working prototype with a small beginner course, rather than a complete language curriculum.

My contribution

A personal Spanish-learning prototype developed with AI coding assistance, covering listening, written recall, spaced review, and guided conversation practice.

SoloPoco beginner Spanish lesson with listening and comprehension exercisesTry SoloPoco
6-2Weightlifting & running
6-3Guitar & music
6-5Cooking
6-6Movies

Interested in machine learning, data, and software engineering opportunities.

Mehdi wearing a graduation cap, with his neck visible and googly eyes