~/wayne $ whoami

Wayne Wen

Backend & AI engineer. Computer Science at the University of Waterloo.

Currently a Software Development Engineer Intern at Amazon, Prime Tech

About

I'm a backend-focused engineer who builds AI systems that have to work in production: observability for large services, security for crypto wallets and agents that take on real tasks.

I study Computer Science at the University of Waterloo with an AI specialization and a minor in Combinatorics & Optimization. Outside work, I usually end up teaching a program to play a game. My 2048 agent now reaches the 8192 tile.

Waterloo CS '28AI specialization, C&O minor
5 internshipsAmazon, BitGo and more
Backend & AIGo, Python, Java, TypeScript
Vancouver, BCCurrent location

Experience

Amazon

Software Development Engineer Intern, Prime Tech · Vancouver, BC

  • Built and deployed a multi-agent incident analysis system that correlates CloudWatch telemetry across AWS services, surfaces related past incidents and gives on-call engineers evidence-backed reports.
  • Caught 10+ production degradations in one week, confirmed by on-call engineers. Several had no alarm at all, and one SEV2 outage was flagged 46 minutes before its alarm. Tuned the anomaly detection on a 15M-datapoint backtest, cutting false positives 7x.
  • Scaled ingestion from 5 metrics to 20K+ time series by streaming CloudWatch pages into SQLite under rate-limited concurrent requests, cutting fetch time 2.8x and memory 6.7x, then made the analysis 10x faster with byte-identical results.
  • Designed a rollout guardrail that compares one-minute live telemetry against pre-rollout baselines and warns before alarm thresholds trip, validated by replaying real rollouts, including a load test that peaked at 52K errors per minute.

BitGo

Software Engineer Intern, Wallet Core · Palo Alto, CA

  • Resolved a transaction-signing issue across 30+ EVM-family coins, cutting withdrawal errors on mainnet by 95%.
  • Engineered WebAuthn intent attestation in Node.js and TypeScript, binding transaction intent to passkey signatures and re-verifying it throughout multisig and MPC withdrawal flows.
  • Secured 280K+ transactions against in-flight payload tampering, database record tampering and replay attacks with origin checks, intent encryption and verification at every identified single point of failure.
  • Built Grafana dashboards for attack vectors, signature failures and false positives, and shaped non-blocking rollout safeguards.

Emailgistics

Senior Software Engineer Intern · Oakville, ON

  • Developed Go microservice features for email assignment rules, configurable alerts and new report types.
  • Parallelized report generation and settings loading with concurrency, and cut alert latency by 20–25% with flattened MongoDB collections and compound indexes.
  • Designed internal REST APIs for alert configuration and status with sub-100ms responses.

Supernova MGU

AI Software Engineer Intern · Toronto, ON

  • Built a Python RAG agent that scrapes social media asynchronously, analyzes high-performing posts and generates content with 66% higher engagement scores than a non-agent baseline.
  • Implemented fuzzy-matching search with RapidFuzz over scraped posts, reaching 90%+ keyword relevance in retrieval tests.
  • Automated new-broker onboarding with Python, n8n and REST APIs, reducing the onboarding error rate by 92%.

Caribou Contest

Full Stack Developer Intern · Thorold, ON

  • Improved mass-email delivery throughput by about 50% by optimizing algorithms and SQL queries.

Projects

2048-Auto

A CNN value-network agent that plays 2048, trained with TD(λ) reinforcement learning, experience replay, symmetry augmentation and expectimax search.

52%games reaching 8192
94.8%games reaching 4096
20Ktraining episodes
PythonTensorFlowKerasFlask

fantasy_cracker

An autopilot for 9-category fantasy basketball that combines ML projections, Claude's reading of the news and 2,000-run matchup simulations to pick the best moves.

PythonClaudeDocker

WebWhisper

A browser side-panel AI agent for YouTube, Reddit, LinkedIn and Instagram, with bring-your-own-key support for five model providers.

TypeScriptGo

Sidechat

A chat UI with side questions that see the whole conversation without changing it, plus conversation forking and rewind.

ReactTypeScriptFastAPI

Chamber Claw 3000

A text-based dungeon crawler in modern C++, designed for low coupling and high cohesion.

C++OOP

Skills

Languages

GoPythonJavaTypeScriptJavaScriptC/C++RustSQLShell

Frameworks & APIs

Spring BootGinFastAPIFlaskReactVue.jsNode.jsRESTSSE

Cloud & DevOps

AWSCloudWatchDynamoDBBedrockDockerKubernetesKafkaJenkinsGrafana

Data & AI

PostgreSQLMongoDBRedisChromaDBTensorFlowKerasLangChainRAGMulti-agent systems

Education

University of Waterloo

Bachelor of Computer Science, AI specialization, Combinatorics & Optimization minor · 2023 – 2028

Contact

Open to conversations about backend systems, AI agents and internship opportunities.