Computer Engineering student at the University of Waterloo and AI Engineering Intern at Presto Phoenix. Building agentic voice systems, machine learning, and full-stack products through work at Presto, BMO, and WATonomous.
A hold-to-talk AI persona of me — Whisper fine-tuned with LoRA and exported to CTranslate2 for real-time int8 transcription, driving a LangGraph agent grounded with RAG over personal data in ChromaDB, with prompt-injection guardrails gated by a CI eval harness.
PyTorch / LangGraph / Whisper / RAG / ChromaDB / HuggingFace
↳ Demo Recording
A 36-hour agentic repair assistant that diagnoses broken household items, grounds replacement-part recommendations in live product catalogs with RAG, and takes in-stock Shopify inventory through checkout.
Python / Agentic AI / RAG / Shopify Storefront API
A collaborative AI study platform that turns lip-read videos into notes and quizzes, built during Hack the North and awarded Best Use of Auth0.
Python / React / Auth0 / MongoDB / OpenAI / Cohere
A real-world ML and hardware build that classifies recycling versus garbage and physically sorts it using an Arduino-driven mechanism.
Python / TensorFlow / Arduino / Computer Vision
An interactive React site for FIFA World Cup statistics with dynamic filtering and real-time graph updates. A pandas and Matplotlib backend pipeline queries data and generates plots, reducing manual processing time by 95%.
React / Python / SQL / Matplotlib / pandas
A teaching-focused robotics site plus robot programming work for Team 854, created to help newer programmers ramp up faster.
Java / WPILIB / Robotics / React
A first-place hackathon rover that navigated a maze autonomously, then switched into Bluetooth control to scan barcodes.
Arduino / C++ / Sensors / Autonomy
↳ Click a row to expand
Working on agentic voice ordering for drive-thru deployments in San Mateo, California: streaming ASR, custom NLU and LLM orchestration, menu grounding over POS integrations, and TTS. Focused on order accuracy (NIR), per-turn latency, failure triage, and release evaluation coverage.
Led a file monitoring system using DQL in Dynatrace and Python and JSON in Ansible to track file presence, metadata, and SLA compliance, cutting manual escalation time by 60%. Helped build a TSYS dashboard for data flows, AWS API and MQ volumes, and consumer usage with AWS Lambda CloudWatch logs.
Trained reinforcement-learning locomotion and imitation-learning manipulation policies in Isaac Lab and PyTorch. Fine-tuned YOLOv11 for competition-specific objects, achieving 95% accuracy in the robotics simulation pipeline, and used WATcloud’s Slurm-managed GPU cluster for distributed training.
Education
Bachelor of Applied Science in Computer Engineering · GPA: 3.3
Certification
Elasticsearch, EC2, ECS, CloudWatch, CloudFront, and S3.
Training reinforcement-learning and imitation-learning policies, plus models for object detection, transcription workflows, and applied computer vision.
Building grounded agents and voice pipelines with streaming ASR, LLM orchestration, RAG, and release evaluations — from drive-thru ordering to repair-part checkout.
Building interactive products and data workflows with React, Next.js, Python, SQL, Supabase, and AWS.
YOLOv11 competition object accuracy in the WATonomous pipeline
Manual data processing time reduced in the FIFA World Cup graphs pipeline
Manual escalation time reduced through file monitor automation at BMO