Calvin Dobbs

Software engineer and startup co-founder building backend, cloud, and edge AI systems.
Computer Science and Mathematics at the University of Toronto, graduating May 2028.

Selected work

Agentiiv Model Context Protocol Gateway

Student-run AI team / Jan to Apr 2026

A secure gateway for connecting AI agents to external tools, with authentication, role-based access, audit logging, rate limiting, and monitoring.

My work: led eight developers, translated partner requirements into Jira workstreams, and delivered the agreed MCP gateway functionality on schedule. I established the FastAPI and PostgreSQL foundation and implemented the AWS deployment for nine containerized services.

The AWS deployment routes traffic through CloudFront and an ALB to nine services on ECS Fargate, backed by RDS PostgreSQL, private networking, service discovery, and secrets management.
Leadership
8 developers
Deployment
9 services on AWS
Code
Private under partner agreement

AfterShock

Autonomous browser QA and repair / Hack the North 2026

Turns a GitHub push into an evidence-backed browser QA run: it derives assertions from the diff, reproduces regressions, generates a patch, replays the failing journey, and publishes a repair PR only after the replay passes.

My work: built the typed contracts and orchestration that connect each pipeline stage, instrumented Stagehand and Browserbase runs with durable telemetry and screenshot replay, and implemented diagnosis, Codex-backed patch generation, deterministic re-verification, and verified GitHub publishing.

In a live rehearsal, the system found two failures, generated and deployed a patch, replayed the failed journey successfully, and opened a verified repair PR in approximately three minutes.

Stack: TypeScript, React, Next.js, Stagehand, Browserbase, SSE, GitHub API

Crafter

Minecraft-to-robot building system / Battle of the Schools 2026

Converts structures designed in Minecraft into machine-readable blueprints for a robot that can locate physical blocks and reproduce the build in the real world.

My work: built a Minecraft mod that turns nearby structures into robot-readable blueprints, plus the core robot-control software for live perception, autonomous surveying, block placement, navigation, offline simulation, and operator debugging.

Stack: Python, Java, Fabric, computer vision, robot control, HTML/CSS/JavaScript

The Mind Museum

Generative 3D learning environment / GenAI Genesis 2026

My work: built the initial Flask and Docker backend and asynchronous SSE world-generation pipeline, then integrated Gemini-driven artifact generation, image-to-3D creation, and interactive React and Three.js museum experiences.

Experience

Co-Founder and Software Engineer

Greywing Technologies / Aug 2026 to present

  • Co-founded Greywing and lead software development across offline edge services, acoustic and visual detection, operator alerts, and hardware integration.

Engineering Project Manager

University of Toronto Machine Intelligence Student Team (UTMIST) / Aug 2026 to present

  • Oversee two internal machine-learning projects and one company-sponsored project, selecting and mentoring team leads and guiding technical scope, architecture, and delivery.

Development Team Lead

University of Toronto Machine Intelligence Student Team (UTMIST) / Jan to Apr 2026

  • Led eight developers on a semester-long project for Agentiiv, translating partner requirements into Jira workstreams and delivering the agreed MCP gateway functionality on schedule.

Backend Development Intern

Asigra / May to Aug 2025

  • Traced several QA failures to concurrent Lambda writes using CloudWatch and targeted tests, then redesigned DynamoDB storage to prevent lost updates.
  • Moved frequently updated collections into dedicated DynamoDB tables and refactored Java Lambdas to isolate independent workflows.

Full Stack Web Development Intern

Enginess / Jan to Jun 2023, May to Aug 2024

  • Built a C#/.NET module for Advantage CSP that transformed Google Analytics data into date-filtered traffic and geographic dashboards in the content management system's admin interface.
  • Built the Node.js and SQL Server foundation for Bell Canada's B2B eOrdering modernization, including authentication, sessions, convention-based routing, a bilingual Handlebars UI, and a Webpack pipeline.

Education

University of Toronto

BSc, double major in Computer Science and Mathematics / expected May 2028

cGPA 3.88/4.00. Dean's List, 2023-2026. First place in the Model Context Protocol Trust Gateway Case Competition at EigenAI. Reuben Wells Leonard Scholarship and Connie P. Dickson Scholarship.

Completed
Machine Learning, Probability and Statistics, Algorithms
In progress
Deep Learning, Probabilistic Learning and Reasoning, Natural Language Computing

Tools

Languages
Python, Java, C#, TypeScript, JavaScript, SQL
Frameworks and APIs
FastAPI, Quarkus, .NET/ASP.NET, Node.js, React, Handlebars, REST, Model Context Protocol
Cloud and data
AWS ECS/Fargate, ECR, RDS, Lambda, DynamoDB, CloudWatch, Docker, PostgreSQL, SQL Server
Tools
Git, Jira, Prometheus, Grafana, Webpack, Figma