About Me

👋🏾 Hi! I’m Tawana, a Computer Science student at Carleton University (Honours, Co-op) with a minor in Business and a 3.95 GPA. I like building software at the boundary between machine learning and the production systems that have to run it.

đź’» I spent this past summer at Shopify on Privacy Engineering, working on GDPR/CCPA data subject requests at petabyte scale. Before that I was at Microsoft on the IDNA edge networking team, building automated root-cause analysis for their global reverse proxy, and at Raven Connected on computer vision and edge ML for connected vehicles. Graduating May 2027.

Tawana Hondonga

Experience

  • Software Engineer Intern @ Shopify — Privacy Engineering

    - Decommissioned the legacy PII redaction engine end-to-end across Rails, MySQL, BigQuery, and Terraform in dependency order — 17,200+ lines deleted with zero regressions, cutting ~56% of a $120–140K/month bill.
    - Halted unbounded growth on a rate-limited deletion queue after the backlog hit 5.77M jobs / 29.6 GB of Redis, shipping producer-side back-pressure gated on consumer depth and a drain-rate metric.
    - Migrated live schemas on a Rails/MySQL service under continuous production traffic, staging each change across dual-write, backfill, and cutover deploys so reads and writes stayed correct at every intermediate state.

    Duration: May 2026 - August 2026

  • AI Systems Engineer Intern @ Raven Connected

    - Attacked demographic and lighting bias that made the cabin-facing model fail out-of-distribution, building a synthetic data pipeline that captions images with Vision LLMs, then generates controlled variations isolating the target label.
    - Trained classification heads on a frozen DINOv3 backbone and distilled a MobileNetV2 student sized for on-device inference on Raven+ edge hardware.
    - Built an automated cabin-mapping labelling pipeline chaining monocular depth (Depth Anything V3) with instance segmentation (SAM 3) into occupancy grids that auto-assign driver and passenger labels by seat position, replacing manual annotation.

    Duration: January 2026 - April 2026

  • Software Engineer Intern @ Microsoft — IDNA (Edge Networking)

    - Built an automated root-cause analysis tool for Microsoft’s company-wide reverse proxy (50+ services, all global regions, trillions of requests/day), cutting root-cause identification from hours to under 5 minutes in backtests on historical incidents.
    - Designed a multi-tier diagnostic engine in Python that parallelizes anomaly detection across 100+ metrics with per-metric fault isolation, narrowing failures from the proxy layer to a single service deployment before an LLM agent issues a verdict.
    - Compiled natural-language troubleshooting guides into executable diagnostic workflows via a LangChain multi-agent prototype, letting on-call engineers extend automated recovery coverage without writing code.

    Duration: May 2025 - July 2025

  • AI/ML Engineer Intern @ Raven Connected

    - Shipped a semantic construction-zone mapping pipeline chaining monocular depth with segmentation across 40 km of the QEW Innovation Corridor, now in production for roadway-safety enforcement in Toronto and Atlanta.
    - Delivered the company’s first speed-limit sign detection, curating and labelling a 15,000-image dataset and fine-tuning to 0.88 mAP@[.5:.95] — now in production for real-time driver assistance.
    - Built the company’s first seat-belt compliance model, a two-stage detector/classifier pipeline reaching 97% accuracy and 0.99 F1 with zero regression on existing safety-event classes.

    Duration: September 2024 - April 2025

  • Teaching Assistant @ Carleton University

    - Led technical tutorials and code reviews for COMP 1405 (700+ students), mentoring students through debugging and foundational software architecture.

    Duration: September 2023 - April 2024

Tawana Hondonga

Projects

Skej Project

Skej (uOttaHack 2nd Place)

AI-powered course scheduler generating conflict-free timetables in <5 mins.
(Typescript, HTML, CSS, Python)

Tawana Hondonga

DataPrepML

ML pipeline & Data preprocessing tool.
(Python)

Tawana Hondonga

EchoGPT

Smart AI Assistant.
(OpenAI API, React, Next JS, Firebase, Vercel)

Tawana Hondonga

iPantry

AI powered pantry tracker.
(React, NextJS, Firebase, Vercel, GCP, Gemini API)

Tawana Hondonga

iMedify

Integrated media consumption platform.
(HTML/CSS/JavaScript, Node, Express, SQLite)

Tawana Hondonga

Personal Website

You're already there!
(HTML/CSS)

Tawana Hondonga

Phantom Pursuit

Multithreaded simulation using POSIX threads & semaphores.
(C, Linux, Makefiles, Valgrind)

Tawana Hondonga

Migraine Classification Neural Network

Classifies seven types of migraines with 93% accuracy.
(Python, TensorFlow, Pandas, Scikit-learn)

Even more...

đź’» Technical skills:
Languages: Python, C/C++, Java, Ruby, TypeScript/JavaScript, SQL, KQL, Bash
ML & AI: PyTorch, TensorFlow, scikit-learn, OpenCV, Hugging Face, LangChain, ONNX/TFLite, distillation, fine-tuning
Backend & Data: GraphQL, Redis/Sidekiq, MySQL, Ruby on Rails, Node.js, Next.js, BigQuery, MongoDB, pandas
Infrastructure: Azure, AWS (Certified Cloud Practitioner), GCP, Docker, Terraform, Git, CI/CD, Linux

🎓 Awards & Honors:
● President’s Faculty Scholarship ($16k)
â—Ź Award of Excellence ($5k)
● Dean’s Honor List (2022-2025)
â—Ź Vice President & Co-founder @ ColorStackCU

Tawana Hondonga