Secret Route
← Return
STAND USER · スタンド使い
ARCANA: THE MAGICIANStand User

KEN WU

吴锵皓
SOFTWARE ENGINEERML RESEARCHERCS '26
LOCATION
Toronto, ON
BASE
University of Waterloo
SEEKING
Fall 2026 / New Grad
Stand Name
DEEP PURPLE
Stand Stats
DESTRUCTIVE POWER
B
SPEED
A
RANGE
A
DURABILITY
B
PRECISION
A
DEV POTENTIAL
∞
Battle History · 戦歴
Chapter 01 · ML · ATTENTION & SYMBOLIC REGRESSION

University of Waterloo

With Prof. Ali Ghodsi & Amin Ravanbakhsh
Sep 2024 — Present
Waterloo, Canada · Remote
Undergraduate Research Assistant
▶Benchmarked symbolic regression at dataset scale, holding R² ≥ 0.99 whenever fits stayed numerically stable.
▶Fine-tuned Symbolic GPT variants for roughly 19 percentage points higher in-domain accuracy.
▶Ablated tokenizer and Point-Net configurations to balance R² against overall model complexity.
▶Refined inference loops to reduce MSE and MRE consistently across standard benchmark suites.
PythonPyTorchTransformers
Chapter 02 · UNSUPERVISED LEARNING

Lancaster University

With Prof. Plamen Angelov
Jan 2025 — May 2026
Lancaster, UK · On-site
Undergraduate Researcher
▶Recursive ReSil / ReSilC in Python: O(1) key updates and PAMSil up to 85.6% faster on CIFAR-100 at equal quality.
▶Optimized R-Means centroid updates for 17–24% faster runs than K-Means on CIFAR-10/100, MNIST, and Fashion-MNIST.
▶Built a NumPy / scikit-learn pipeline benchmarking recursive versus flat clustering across 8+ datasets with 10-run averages.
▶Tracked silhouette, inertia, and wall-clock time each run so speed-quality comparisons stayed fair.
PythonNumPyscikit-learn
Chapter 03 · ML PLATFORM

Lyft

May 2026 — Present
Toronto, Canada
Machine Learning Software Engineer Intern
▶Working on rider experience related ML model training, evaluation and observability development.
Chapter 04 · AGENTIC LAGRANGIAN EXTRACTION

Fermilab

Google Summer of Code · ML4SCI · University of Alabama
May 2026 — Aug 2026
Remote · United States
Open Source Developer
▶Open-source contributor via Google Summer of Code 2026 with ML4SCI at Fermilab, collaborating with mentors at the University of Alabama.
▶Project focus: Agentic Lagrangian Extraction — details on the GSoC project page.
Python
Chapter 05 · 5G NETWORK AGENT

Nokia

Jul 2025 — Dec 2025
Ottawa, Canada
Software Engineer Intern
▶Built LoRA fine-tuning pipeline for Qwen generating Camunda BPMN XML with GGUF quantization
▶Designed evaluation framework revealing overfitting from NCSC-only data vs. base models
▶Resolved GPU memory and Triton issues and shipped reusable training tooling for future experiments
PythonPyTorchUnslothTransformer
Chapter 06 · INSURANCE ANALYSIS

TD Bank

Jun 2025 — Aug 2025
Toronto, Canada
Data Scientist Intern
▶Replatformed pipelines for 1.5M+ rows of data and reproduced a 154,340 row deliverable with 100% parity
▶Automated 30+ minutes of manual ingestion per cycle, cutting QA time by 80% via a parity harness
▶Migrated on-prem pipeline to Azure fully automating runs and saving 2–4 hours/week of manual execution
PythonSQLPySparkPandasDatabricks
Chapter 07 · TEACHING

Stanford University

Apr 2025 — Jun 2025
Stanford, United States
Student Instructor
▶Taught Stanford’s Code in Place CS106A course to students globally, taken by 40,000+ students
▶Educated students in Python leveraging beginner friendly libraries including Stanford’s Karel and Tkinter
PythonKarelTkinter
Chapter 08 · LLM AGENT

August

Sept 2024 — Dec 2024
New York, United States
Software Engineer Intern
▶Handled 2,000+ requests/min by deploying 10+ API endpoints using FastAPI, AWS, and Supabase
▶Optimized evaluation cycles for 15+ LLM Agents with a round-robin multi-agent and scoring framework
▶Led the end-to-end development of a multi-agent RAG pipeline powered by LLM-as-Judge strategies
PythonLangGraphFastAPIAWS
Chapter 09 · SUPER RESOLUTION

hum.ai

Formerly Coastal Carbon
Jul 2024 — Sep 2024
Kitchener, Canada
Machine Learning Engineer Intern
▶Benchmarked SOTA super-resolution models (e.g. ESRGAN, StableSR) through PyTorch pipelines
▶Built automated benchmarking pipelines in Python to evaluate multiple models efficiently
▶Visualized model performance with Matplotlib and Seaborn in Jupyter on SageMaker for analysis
▶Managed experiment infrastructure on AWS S3 and EC2 for scalable fine-tuning and evaluation
PythonPyTorchAWSSageMakerJupyter
Chapter 10 · DOCUMENT QA

Health Canada

Apr 2024 — Aug 2024
Ottawa, Canada
Machine Learning Engineer Intern
▶Built a document QA system using Llama3 7B and ChromaDB for OECD report search and summarization
▶Increased response and semantic accuracy by ~20% using query transformation and contextual memory
PythonAzureLangChainStreamlit
Chapter 11 · OPERATIONS & AUTOMATION

Saputo

Jan 2024 — Apr 2024
Georgetown, Canada
Data Analyst Intern
▶Developed TypeScript Office Scripts in Excel that eliminated ~8 hours/week of manual open-order updates
▶Automated weekly workflows for 1000+ Nestlé products, avoiding 20,000+ manual data entries
▶Used the Gemini API to automate competitor research across 200+ brands
▶Ran weekly statistical analysis and EDA in VBA across 2000+ major products and 200+ miscellaneous SKUs
TypeScriptExcelVBAGemini API
Chapter 12 · RESUME PARSING

Respan

Y Combinator W24 · Formerly Keywords AI
Mar 2023 — Jun 2023
New York, United States
Software Engineer Intern
▶Parsed 1,000+ resumes with a spaCy-based NER pipeline to extract structured recruiter data
▶Reduced response delay by 98% through integration of SQLite-based result caching into the parsing engine
PythonspaCySQLite
Chapter 13 · ENTITY RECOGNITION

Intapp

Formerly delphai
Jul 2022 — Sept 2022
Berlin, Germany
Machine Learning Engineer Intern
▶Boosted recall by 20% through improved entity labeling workflows and language-specific training sets
▶Achieved 70% recall by fine-tuning spaCy models and optimizing hyperparameters via WanDB on Azure
PythonspaCyW&BBS4
Operations · 作戦
Op · 01

OpenComposer

Mini-scale end-to-end reproduction of the Composer 2 training pipeline on a single NVIDIA GH200: MoE CPT, MTP self-distillation, Dr. GRPO + k1 KL, and behavioral auxiliary rewards.

PythonPyTorchMoEOpenRLHF
↗ REPO
Op · 02

AppGammon

Online backgammon with a React Native/Expo client, Hono backend, and shared gameplay rules in a TypeScript monorepo.

TypeScriptReact NativeExpoHono
↗ REPO
Op · 03

Ding-Bot

Chess engine combining GATEAU-style Graph Attention Networks with contrastive latent-space search

PythonTypeScriptGraph Neural Networks
↗ REPO↗ DEMO
Op · 04

PokerMon

Deep Counterfactual Regret Minimization (Deep CFR) for 6-player No-Limit Texas Hold'em

PythonDeep CFRGame Theory
↗ REPO↗ DEMO
Op · 05

LeaseEase

Streamlit app demystifying Canada’s Residential Tenancy Act with LLM + RAG, plain-language guidance, and auto-generated forms (T1, N7) for tenants navigating the housing crisis.

PythonStreamlitOpenAICohereChromaDB
↗ DEMO
Op · 06

MedChat

Assistant for clinical Q&A: Cohere Classify routes intent to a brain-tumor CNN or RAG over 1000+ WebMD pages with streamed answers in Streamlit.

PythonCohereTensorFlowStreamlit
↗ REPO
Op · 07

DirectU

Full-stack planner matching career goals and free-text course preferences to UWFlow reviews via Cohere, assembling a personalized four-year roadmap (React, Flask, MongoDB).

ReactFlaskMongoDBCohere
↗ DEMO
Op · 08

LeGM-Lab

AI-powered NBA take analyzer that fact-checks basketball opinions with real stats and roasts bad takes on X

PythonClaude APIFastAPIX API
↗ REPO↗ DEMO
Op · 09

FlightCal

Fetches flight info and exports it directly to Google Calendar or as an .ics file for any calendar app

TypeScriptNext.jsGoogle Calendar API
↗ REPO↗ DEMO
Stand Abilities · 能力
Languages
PythonSQLCC++TypeScriptJavaScriptHTMLCSSRRacketBash
Technologies & Cloud
PySparkNumPyPandasspaCyMongoDBSupabasePostgreSQLAWSDockerGit
Libraries & Frameworks
PyTorchLangChainFlaskFastAPITensorFlowKerasCUDAReactNext.jsTailwind
Contact · 接触

Your story has not ended.

— Connect with the Stand User

GitHubLinkedInEmailResumeX
CONFIDENTIAL · 極秘
〜 END OF RECORD · 記録終了 〜
秘