Portrait of Jemima Silaen

Jemima Silaen

cs + stats @ uoft

About

hi there! i’m Jemima (but you can call me Mima) :)

i’m currently studying computer science and statistics at the university of toronto. my interests span applied ML, NLP, GPU programming, cognitive science, and linguistics.

growing up in jakarta and finding my way to toronto, i’ve come to love the feeling of being somewhere unfamiliar. i like throwing myself into things i know little about, figuring things out along the way, and discovering how much there is to learn when you stop being afraid of starting from scratch.

outside of school and work, i looove to dance, play piano, edit videos, discover new music, and watch films! some of my favorites are All That Jazz, The Sound of Music, Lady Bird and Synecdoche, New York.

if you’d like to chat about anything, email me!

Experience

Internships

  1. Data Scientist · Intact

    September 2026 – December 2026

    Building agentic pipelines for automated feature discovery and evaluation in insurance pricing models.

  2. Data Scientist · NTT Data

    June 2026 – September 2026

    Developed a model explainability platform for analyzing and monitoring credit risk models.

  3. Cybersecurity Intern · Ernst & Young

    June 2025 – August 2025

    Worked on digital identity and cybersecurity risk analytics, synthesizing findings into Power BI dashboards for financial institutions.

Clubs

Developer

University of Toronto Machine Intelligence Student Team

  • Sparse Attention
  • LLM Inference
  • Long-Context Modeling
  • Transformers

Deep Learning Engineer

University of Toronto Formula Racing

  • Computer Vision
  • Active Perception
  • Imitation Learning
  • Knowledge Distillation
  • DAgger

Projects

  • TinyINR

    A flow-matching ML framework built from scratch in C++/CUDA, with custom GPU kernels, manual backprop, and adaptive sampling.

    GitHub
  • GradientMesh

    A federated learning runtime in C that trains across Linux and embedded workers over custom communication protocols.

    GitHub
  • Food Map

    A Next.js and Mapbox GL map for finding, rating, and sharing food trucks, restaurants, and cafés around UofT St. George.

    GitHub
  • Warden

    A multi-agent system built with Google ADK and Next.js that flags supply chain vulnerabilities, shown in an interactive Three.js stockroom.

    GitHub
  • Memory Palace

    A dementia-care platform that turns memories into explorable 3D worlds with World Labs and Three.js, plus Whisper speech recognition and GPT-guided conversations.

    GitHub
  • GigIT

    A mortgage verification platform that automates document processing and validation with FastAPI and Gemini 2.5 Pro.

    GitHub
  • Healthz

    A Java and JavaFX nutrition and fitness tracker with food search and activity tracking, built on clean architecture.

    GitHub
  • Politik Yuk

    A political news platform built with LangGraph, Cohere, and RAG that delivers source-grounded summaries with citations.

    GitHub

Skills

Languages

  • Python
  • C/C++
  • TypeScript/JavaScript
  • Java
  • SQL
  • Bash/Shell

ML & Data

  • PySpark
  • PyTorch
  • CUDA
  • ONNX
  • scikit-learn
  • XGBoost
  • Transformers
  • NumPy
  • pandas
  • BERTopic

Developer Tools

  • AWS
  • Databricks
  • GCP
  • FastAPI
  • PostgreSQL
  • Redis
  • Docker
  • Linux/Unix
  • Git
  • CMake
  • SLURM
  • Next.js
  • React
  • Flask
  • Supabase
  • Pydantic
  • Agile