Skip to content
pv

Software developerCalgary, Alberta

Parmeet
Virdi

I design and build backend systems, distributed architectures, and full-stack applications.

My work spans backend systems, distributed computing, React interfaces, and data integration. My projects explore scalable services and applied machine learning.

Currently a software developer at Snap Remedy.

Professional experience
Backend systems, data integration & frontend development Based in Calgary, Alberta

Professional experience

Full experience
Apr. 2026 – PresentSoftware Developer

Snap Remedy

ERP data pipelines and configurable SKU ranking

I developed the Django backend and React frontend, including the relational data model and the interface for configuring comparisons and inspecting results.

ETL pipelines map incoming records into local storage, with checkpoints supporting failed-page replay.

Python / Django / React / SQL

More about my work

Data to results

  1. Data sources

    ERP records and manual inputs

  2. Django backend

    Ingestion, relational models, scoring, and APIs

  3. React frontend

    Configure comparisons and explore results

PRO-Formance Intermodal

Jan. 2026 – Apr. 2026

Software Intern

Sole developer of a full-stack internal system for workflow automation, training scheduling, grant processing, and operational analytics.

View role

Phillys

Sep. 2025 – Nov. 2025

Web Developer

Rebuilt a legacy website with PHP and JavaScript, delivering responsive layouts, web deployment, and post-launch production support.

View role

Selected projects

View all projects

Discover a host. Relay between peers.

Client lookupAddressing primary
Local broadcast, peer forwarding, and seen-ID suppression. Selected relay path shown.
Explore the interactive architecture

Distributed systems / Java NIO

Distributed Chat System

I built IRC-HAT with custom TCP networking, configurable gossip replication, and primary-led server discovery backed by replicated registries.

Java NIO / Replication / Bully leader election

Read project

Synthetic occlusion, real evaluation

3,312clean pet images
3,000synthetic composites

6,312 training images Whole-pet boxes retained

YOLOv8n / EfficientDet-D0

Clean test
3,669 images
Real occlusion
74 images / 90 boxes
Clean vs. combined training. Real-occlusion transfer was mixed across detectors.

Computer vision / PyTorch

Pet detection under occlusion

I built a synthetic occlusion pipeline and detector adapters to test how training on household-object composites transfers to naturally occluded pets.

74 real images / YOLOv8n and EfficientDet-D0

Read study

Language implementation

J-- to MIPS compiler

I built the grammar, semantic analysis, and a MIPS backend. Resolved declarations connect source expressions to target storage.

Source to assembly

J-- sourceprinti(42);
MIPS excerptli $v0, 42 ⋮ jal printi

Parse → Resolve names & types → Emit

Illustrative excerpt; stack setup and runtime omitted. Calls use a custom stack convention.

Four semantic passes check names, types, and structural rules before assembly emission.

Read project

More projects

Genetic hyperparameter search Python / scikit-learnSoccer scheduling Python / Constraint searchRestaurant reservations React / Express / MySQLBare-metal controller C / GPIO / UARTGreenhouse simulation Java / SwingFamily streaming interface React / MUI

Outside work

I enjoy hiking around Kananaskis and exploring Alberta. Being based in Calgary makes it easy to get out to the mountains. There’s still plenty of the province I’d like to see.

A forested valley with lakes, snow-covered mountains, and a clear blue sky, viewed from a hillside.
A rocky mountain reflected in a lake, with snow on its slopes and low cloud along the shoreline.