Satyam Bhanot · Computer Science Honours

Projects that show my learning.

CS Honours graduate from the University of Manitoba with a Math and Statistics minor. I build backend, data, systems, and ML work by defining the question, testing a version, debugging it, and writing down what I learned.

View Projects Download Resume GitHub

6 Projects what I tried and learned

Java · Python · C primary languages

3.90 GPA Dean's Honour List, Fall 2024

Learning Through Projects

What I have practiced, and which projects taught it.

Backend & Apps

User actions, APIs, and state that stay predictable.

JavaPythonFastAPINode.js

Data Work

Cleaning inputs, joining sources, and checking assumptions.

SQLPostgreSQLREST APIsJupyter

Machine Learning

Comparing models and reading metrics carefully.

scikit-learnPyTorchXGBoostMetrics

Systems Programming

Debugging memory, files, processes, and edge cases.

CLinuxgdbValgrind

Learning Map

Each project taught one main lesson.

Comparing models

Fraud Detection · Waldo CNN I learned that the metric changes the answer.

Working with data

Winnipeg City Analysis · Fraud Dataset I learned to check assumptions before trusting a join or score.

Understanding systems

ExFAT Shell · Threads Library I learned how small rules create visible behavior.

Building usable flows

PetObject · Support Work I learned that clear structure and communication matter as much as code.

Main Projects

Projects, what I tried, and what I learned.

Machine Learning

Credit Card Fraud Detection

Tested six models on a dataset where fraud cases were rare.

LearnedAccuracy can hide weak detection

Pythonscikit-learnXGBoostLightGBMSMOTE

Learning Notes

Fraud Detection

Question

How do you evaluate a model when the rare class matters most?

What I Tried

Compared six classifiers with SMOTE, class weighting, and different metrics.

What I Learned

Accuracy can be misleading; precision and recall tell the real story.

GitHub Live demo not available

Computer Vision

Where's Waldo? CNN Detection

Compared YOLOv8, RetinaNet, SSD MobileNetV2, and a custom CNN across annotated and synthetic image data.

LearnedF1 and precision answer different questions

PythonPyTorchYOLOv8RetinaNetSSD

COMP 4360 · Machine Learning · University of Manitoba

Learning Notes

Waldo Detection

Question

Which detector handles small, easy-to-miss objects best?

What I Tried

Compared YOLOv8, RetinaNet, SSD MobileNetV2, and a custom CNN.

What I Learned

A model with better F1 and a model with better precision can be useful in different ways.

GitHub Live demo not available

Systems Programming

ExFAT Shell

Built a C shell for ExFAT file-system interaction with UNIX-style commands, redirection, and piping.

LearnedSmall parsing rules shape the whole tool

CLinuxFile SystemsProcesses

COMP 3430 · Operating Systems · University of Manitoba

Learning Notes

ExFAT Shell

Question

What has to happen behind a simple shell command?

What I Tried

Implemented command handling, process execution, redirection, and piping.

What I Learned

Parsing and error cases are where simple tools become reliable.

GitHub Live demo not available

More Projects

Smaller projects that taught useful lessons.

Winnipeg City Analysis

Joined civic datasets from Winnipeg's open data API to compare neighbourhood access and risk signals.

JavaSQLREST API

Learning Notes

Winnipeg Analysis

Question

Can public data answer a neighbourhood comparison question?

What I Tried

Joined transit, parks, hospitals, and incident data from different schemas.

What I Learned

The hard part is deciding what counts as the same place.

GitHub Live demo not available

Concurrency

Cooperative Threads Library

Implemented a pthread-style C thread library with mutexes and multiple scheduling policies.

CSchedulingValgrind

Learning Notes

Threads Library

Question

How do scheduling rules change program behavior?

What I Tried

Built mutex support and FIFO, round-robin, and MLFQ scheduling policies.

What I Learned

Concurrency is easier to reason about when the rules are explicit.

GitHub Live demo not available

Software Engineering

PetObject

Built a team Android app for virtual pet management, validation, editing flows, and tests.

JavaAndroidJUnit

Learning Notes

PetObject

Question

How do you keep a team-built app predictable?

What I Tried

Worked on validation, navigation, edit flows, and tests.

What I Learned

Tests and structure make collaboration less fragile.

GitHub Live demo not available

Experience

Technical work, support, and mentorship.

May 2023 to Present

Tech Support Specialist

Rogers Communications · Remote

50+ weekly incidents across hardware, software, and network support.

TroubleshootingDocumentationEscalationsUser support

Sep 2023 to Aug 2025

Peer Tutor

UofM Science Students' Association · Nimbus Learning

One-on-one and group tutoring across computer science, mathematics, and statistics.

Practice setsConcept breakdownsProblem solvingMentorship

2024 to Present

Student Mentor and Campus Leader

University of Manitoba

Mentorship, orientation support, and student community leadership at UManitoba.

Student transitionsResource navigationBudget trackingCampus leadership

Skills

Grouped by how I use them.

Backend & Apps

APIs, services, and application logic

JavaPythonJavaScript Node.jsFastAPIReact

Systems & Low-level

C-level programming, debugging, memory

CC++Linux gdblldbValgrindMakefile

Data & ML

Pipelines, analysis, and model work

SQLPostgreSQLMongoDB RJupyterscikit-learn

Workflow & Tools

Dev process and collaboration

GitDockerGitHub Figmatestingdocumentation

Education

Computer Science Honours with a Math and Statistics minor.

Bachelor of Science · University of Manitoba · Graduated April 2026

Dean's Honour List · Fall 2024, GPA 3.90

University 1 Honour List · Winter 2021, GPA 3.83

Operating Systems Databases Machine Learning Security Algorithms Statistics

Contact

Open to backend software, data engineering, and ML engineering roles.

New grad, available now. If something here connects with the work you need done, send me a note.

[email protected]

GitHub LinkedIn