open to fall 2026, winter & summer 2027 roles
Mahi Joshi

Mahi
Joshi.

Computer Engineering & Artificial Intelligence student at University of Waterloo building production software, ML systems, and real-time robotics. Targeting SWE, ML, and AI roles on ambitious engineering teams.

software engineering machine learning AI systems robotics
4
internships
$120K
Schulich Scholar
15+
engineers led
See what I build Let's get in touch
✦ play while you browse

🚀 Space Shooter

Arrow keys / WASD to move.
Spacebar to shoot.
Collect ✦ stars for bonus points!

score 0
lives ❤❤❤
best 0
✦ about me

I like building systems that feel useful, fast, and real.

I'm a Computer Engineering & Artificial Intelligence student at the University of Waterloo, focused on SWE, ML, and AI roles. My work sits at the intersection of production software, applied machine learning, and robotics.

I've built across startups, space robotics, capital markets, and embedded systems. I care about clean engineering, measurable impact, and products that survive outside of a demo.

My default mode is simple: understand the problem deeply, ship the smallest useful version, then make it faster, smarter, and easier to use.
currently
AI SWE @ Communitech + robotics research @ UWaterloo
Building agentic AI solutions and real-time multi-robot control systems.
targeting
SWE, ML, and AI internships
Especially big tech, AI product teams, and infra-heavy engineering groups.
strengths
Python, systems, ML, and full-stack product work
Comfortable moving from model logic to backend APIs to user-facing demos.
builder energy
Robots, hackathon apps, embedded systems, and side quests
I like projects where software touches the real world.
outside code
Ballet, baking, running, travel, and a Sudoku book
A little personality, because there's much more to me than what fits on a resume.
✦ What I bring to the table
What I bring to
engineering teams.
A quick scan of the kind of work I want to keep doing: production software, applied AI, real-time systems, and technical ownership.
Ship
I turn rough ideas into working software with clear interfaces, readable code, and user-facing polish.
Learn
I ramp fast across unfamiliar stacks, from ROS 2 and embedded firmware to ML pipelines and cloud tools.
Own
I like ambiguous problems where I can define the path, build the system, and improve it with feedback.
Build
My favorite work combines software with real-world constraints like latency, reliability, and messy data.
✦ interactive
A network that
thinks with you.
Move your cursor across the canvas to fire neurons and watch activations ripple through the layers: input → hidden → output.

This is what I build: systems that process, learn, and respond. The brighter the node, the stronger the activation signal. Output nodes light up with what I can do for your team.
✦ move your cursor over the network to activate neurons
✦ experience

Where I've built
real systems.

A timeline of roles where I have built across AI, robotics, markets, space systems, and production software.

Communitech
UWaterloo
CIBC
MDA Space
Omori
AI Software Engineer
May - Aug 2026
Communitech · Waterloo, Ontario
focus
Agentic AI solutions for real business workflows
style
Startup-paced product building with applied AI
  • Selected to collaborate with high-growth startups and AltaML to design and deploy AI-powered solutions across real-world business applications.
PythonLLMsAWSFull Stack
Undergraduate Research Assistant
Jan - Apr 2026
ERA Robotics Lab · University of Waterloo
focus
Real-time multi-robot coordination and teleoperation
style
Research code with latency and hardware constraints
  • Architecting a sub-100ms real-time human-in-the-loop control stack for multi-robot coordination across 6 RoboMaster platforms via a full ROS 2 pipeline.
  • Enabling wearable-based teleoperation across 5 robots by integrating a SenseGlove interface that maps finger motions to synchronized, low-latency robot actions.
PythonC++ROS 2Robotics
Global Markets Quant Intern
Sep - Dec 2025
CIBC Capital Markets · Toronto, Ontario
focus
Time-series modelling and market regime analysis
style
Data-heavy research with clean validation
  • Delivered 15+ quantitative momentum and regime-detection models using Python-based statistical pipelines on Bloomberg time-series data.
  • Validated a published time-series momentum framework with 100% replication accuracy in R across 3+ asset classes.
PythonRBloomberg TerminalTime Series
Software Engineering Intern
Jan - Apr 2025
MDA Space · Toronto, Ontario
focus
ML-based fault detection for space robotics
style
Simulation data, anomaly detection, and model robustness
  • Optimized Canadarm3 fault detection to 90% precision, 84.2% recall, 87.0% F1 via an LSTM + One-Class SVM anomaly detection pipeline.
  • Automated ~1,200 AGXUnity simulation datasets/day for training data, reducing overfitting under real-world failure conditions.
PythonLSTMSVMAGXUnity
Software Engineering Intern
May - Aug 2024
Omori North America · Toronto, Ontario
focus
Internal tooling for machine configuration workflows
style
Full-stack software with database-backed operations
  • Eliminated 200+ hours/year of operational overhead and cut manual entry errors by 80% by designing and shipping a full-stack machine configuration ordering platform.
  • Designed SQL schemas improving system reliability and reducing data inconsistencies across the configuration workflow.
PythonSQLHTML/CSSFull Stack
✦ projects

Things I've
built & shipped.

🧠
NeuroGuard
Real-time fall detection for at-risk patients using OpenCV + MediaPipe pose estimation + unsupervised ML. End-to-end Python event pipeline with React dashboard and live clinician alerts. Built at Hack the North.
96.9% detection accuracy
⌥ view on github →
PythonOpenCVReact Native
💡
Fundr
Semantic grant-matching platform with ~0.80 cosine similarity across hundreds of records. Swipe UI, Snowflake data integration, FastAPI backend, CI/CD. Shipped in 24 hours at Hack the Valley.
~0.80 avg top-5 similarity
⌥ view on github →
TypeScriptFastAPIGemini API
⚡
Electric Skateboard
Led a 15-member cross-functional team across mechanical, electrical, and firmware from design through system validation. C++ firmware on STM32 with <50ms real-time LED response on 20+ test runs.
<50ms firmware response
C++STM32SolidWorks
🦾
Multi-Robot Teleoperation
Full Python & ROS 2 pipeline for human-in-the-loop teleoperation across 5 RoboMaster robots via SenseGlove wearable. Sub-100ms latency. Ongoing research under faculty supervision at UWaterloo.
<100ms control latency
ROS 2PythonC++
🤖
Pixel - Autonomous Robot
Differential-drive robot on STM32 Nucleo-F401RE that autonomously navigates using HC-SR04 ultrasonic sensing and PWM-based motor control via L298N driver. Modular embedded C firmware with real-time obstacle avoidance.
<100ms real-time control
⌥ view on github →
Embedded CSTM32PWMRobotics
✦ things i live by
Bright little rules
for life.

A few reminders I come back to when I am learning something new, building under pressure, or trying to make an idea real.

🌱
Your only limit is your mindset.
growth first
☀️
Be the sunshine you want to see in the world.
bring good energy
⏳
There is no right time. There is just time, and you choose what to do with it.
start anyway
✨
Progress over perfection.
ship and improve
✦ say hello

Let's build
something great.

Actively looking for SWE, ML, and AI roles at big tech and ambitious engineering teams. Happy to chat about internships, projects, or anything I am building.

✉
email
m6joshi@uwaterloo.ca
in
linkedin
mahi-joshi-ba904b211
gh
github
mahijoshii
mini mahi-bot
hi! i'm your tiny site guide ✨ click me for fun facts about mahi.
!!
💜