Skip to content
dami.peng

Work

Five internships across startups, research and big tech. Mostly backend, infra and the occasional frontend rebuild.

Download résumé (PDF) →

  1. Winter 2027
    Toronto

    Shopify

    Incoming

    Engineering Intern

    • Go
    • Rust
    • Terraform
    • Kubernetes

  2. Jan – Dec 2026
    Toronto · 3 terms

    IBM

    Software Engineer Intern — Skills Network

    Fall 2026 · Infrastructure

    • Cut Redis memory 99% and raised the upload limit from 20MB to 500MB with direct-to-S3 ingestion via pre-signed URLs.
    • Split the TypeScript/Express server and compute-heavy workers into separate Terraform-provisioned Kubernetes deployments — isolated failures, independent scaling, $7K/year saved.
    • Fixed lost WebSocket events for 83% of users by fanning worker emits out to all 6 API replicas over Redis Pub/Sub.

    Summer 2026 · Product + AI

    • Re-architected a 20K-LOC legacy app as a React/TypeScript SPA for 80K monthly users, with merge-blocking OpenAPI contract checks that caught 10+ breaking changes.
    • Cut latency 71% and $10K/year in API costs by migrating LLM/STT models with zero eval regressions.
    • Built realtime multi-agent voice chat on a LangGraph state machine, streaming TTS for sub-1.2s first audio across 92 languages.

    Winter 2026 · Platform

    • Replaced pod log searches with per-job Bull Board logs for 24K monthly BullMQ jobs, surfacing 20+ bugs.
    • Built JSON/HTML/MD translation APIs with AST serialization to preserve structure, cutting hallucinations 40%.
    • Moved 100+ annual portal deployments to Terraform GitOps on GitHub Actions, removing local credentials.
    • TypeScript
    • Kubernetes
    • Terraform
    • Redis
    • AWS S3
    • BullMQ
    • React
    • LangGraph
    • OpenAI
    • ElevenLabs
    • Docker
    • GitOps
    • NGINX
    • WebSockets

  3. Sep – Dec 2025
    Waterloo

    University of Waterloo

    Undergraduate Research Assistant · Accelerated Gradient Descent

    • Built a PyTorch harness to benchmark gradient-descent optimizers on CIFAR-10 classifiers.
    • Extended accelerated optimization algorithms for 10% faster convergence than baseline optimizers.
    • Python
    • PyTorch
    • Pandas
    • Gradient Descent
    • Mathematical Optimization

  4. Jan – Apr 2025
    Toronto

    Civiconnect

    Software Developer Intern · Nurse Recruit

    • Built a Retool dashboard over Supabase Postgres so non-technical clients self-manage 1,000+ user profiles and credit balances.
    • Moved Postgres event listeners to Supabase Realtime broadcasts, cutting job-update latency 66%.
    • Integrated the Stripe API to automate billing across 60+ monthly transactions.
    • Supabase
    • TypeScript
    • AWS S3
    • Stripe API
    • Postgres
    • React