# Jeremy Pinto

header:
  title: Senior Applied Research Scientist
  location: Montreal, Canada
  email: jerpint [at] gmail [dot] com
  phone: phone number upon request

social:
  blog:
    text: "Blog • www.jerpint.io"
    url: https://www.jerpint.io/
    icon: 📝
  github:
    text: github.com/jerpint
    url: https://github.com/jerpint
    icon: fab fa-github
  linkedin:
    text: linkedin.com/in/jeremy-pinto
    url: https://linkedin.com/in/jeremy-pinto
    icon: fab fa-linkedin-in
  hackernews:
    text: HN/jerpint
    url: https://news.ycombinator.com/user?id=jerpint
    icon: fab fa-hacker-news
  huggingface:
    text: HF/jerpint
    url: https://huggingface.co/jerpint
    icon: 🤗
  youtube:
    text: YT/jerpint
    url: https://youtube.com/@jerpint
    icon: fab fa-youtube

Chat with my resume 👉 [jerpint.io/resume](https://www.jerpint.io/resume)

## Summary

Senior applied research scientist with 7+ years of experience modeling, training and deploying production-ready deep learning pipelines.
Led the development of award-winning LLM prompt-hacking research (EMNLP 2023 Best Theme Paper) and contributed to a successful MOOC reaching 8000+ participants.

Specialized in:

- Developing production-ready computer vision and NLP solutions
- Bridging state-of-the-art research with practical business applications
- Implementing and securing large language model workflows
- Leading technical workshops and mentoring ML practitioners

Stuff I build for fun:

- I maintain an ML-focused blog where I post things like [using ControlNet to animate the Game Of Life](https://www.jerpint.io/blog/diffusion-gol/")
- I post tutorials on YouTube like [how to summarize YouTube videos using GenAI](https://www.youtube.com/watch?v=WtMrp2hp94E)

## Key Achievements

- Led HackAPrompt competition with 2800+ participants from 50+ countries, resulting in EMNLP 2023 Best Theme Paper
- Core contributor of Buster, an open-source RAG tool, with 200+ github stars
- Co-authored deep learning course content reaching 8000+ global participants
- Published gender identification algorithm for medical voice analysis, currently integrated in iOS app

## Work Experience

### Senior Applied Research Scientist
#### Mila - Quebec Artificial Intelligence Institute | Sept 2018 - Present

**Key Responsibilities & Achievements:**

- Architected and implemented production-ready deep learning solutions for organizations
- Mentored SMEs through AI adoption programs, resulting in successful implementation of ML solutions in the Canadian AI ecosystem
- Created and delivered hands-on computer vision workshops for 200+ participants
- Supervised MSc. students during their internship
- Co-instructor for ["Deep Learning Essentials"](https://www.edx.org/learn/deep-learning/universite-de-montreal-deep-learning-essentials) MOOC on EdX (8000+ participants), developed and delivered content on Convolutional Neural Networks and ML tools

### Lead Data Scientist
#### Focus21 | May 2017 - June 2018

**Key Achievements:**

- Developed proof-of-concept medical imaging systems for x-ray diagnostics using Mask R-CNN
- Implemented reinforcement learning algorithms for industrial robotics in simulated environments
- Implemented algorithmic trading strategies and analysis tools

## Skills

**AI/ML Technologies:**

- Generative AI: ChatGPT, Claude, LLaMa, cursor/copilot, Hugging Face {transformers, diffusers}
- Deep Learning: PyTorch, Lightning, TensorFlow, Keras, Jax
- ML Tools: Scikit-Learn, pandas, numpy, scipy, WandB, CometML, tensorboard

**Software Development:**

- Languages: Python, Bash, Javascript, Matlab, LaTeX, Markdown
- API & Web: FastAPI, Gradio, Hugging Face
- Data Processing: pandas, NumPy, hf-datasets

**Cloud & Infrastructure:**

- DevOps: Git, CI/CD, Docker, SLURM
- Cloud Platforms: AWS, Azure, Heroku
- Databases: MongoDB, SQLite
- Editors: VSCode, (neo)vim

**MLOps:**

- Experiment Tracking: WandB, CometML, TensorBoard
- Data Version Control: Hugging Face datasets, deeplake, dvc
- Model Serving: TorchServe, ONNX, BentoML, Docker

**Languages:**
- English (Native), French (Native)
- Hebrew (Limited Working), Spanish (Basic)

## Education

### Systems Design Engineering - Vision and Image Processing (VIP) Lab
#### University of Waterloo, MASc. | 2015-2017

- Thesis: "Cancer Classification in Human Brain & Prostate Using Raman Spectroscopy & Machine Learning"
- Led research resulting in 2 peer-reviewed publications
- Trained and deployed urban sound classification models within iOS apps

### Engineering Physics
#### Polytechnique Montréal, B. Eng. | 2010-2014
- Graduated with Distinction
- Awarded DeVinci Profile and International Profile
- Developed novel acoustic camera system for holography validation

## Projects

### HackAPrompt (2023) | [https://paper.hackaprompt.com/]()
- Led development and implementation of global prompt-hacking competition
- Tech Stack: Python, HuggingFace Transformers, PyTorch, FastAPI
- Impact: 2800+ participants, 50+ countries, EMNLP2023 Best Theme Paper
- Surveyed novel methodologies for testing LLM security

### Buster (2022-2024) | [https://github.com/jerpint/buster]()
- Core contributor of open-source RAG tool with citation capabilities and response-monitoring
- Tech Stack: Python, OpenAI, Gradio, Pinecone, MongoDB, Deeplake
- Adopted in research projects at [Mila](https://huggingface.co/spaces/mila-ai4h/AIR) and the [OECD](https://oecd.ai/en/air/)
- 200+ GitHub stars

### VoiceCollab (2021-Present) | [www.voicecollab.us]()

- Lead ML developer for gender-affirming voice care deep-learning models
- Implemented production-grade audio processing pipeline
- Tech Stack: PyTorch, ONNX, Swift, Docker, MongoDB, Firebase
- Peer-reviewed publications

## Selected Publications

- Schulhoff, S, J. Pinto et al. (2023). "Ignore This Title and HackAPrompt: Exposing Systemic Vulnerabilities of LLMs through a Global Scale Prompt Hacking Competition"
EMNLP2023 Best Theme Paper Award

- Bensoussan Y, Pinto J, et al. (2021). "Deep Learning for Voice Gender Identification: Proof-of-concept for Gender-Affirming Voice Care." Laryngoscope

- J. Pinto (2017), "Cancer Classification in Human Brain and Prostate Using Raman Spectroscopy and Machine Learning." MASc. Thesis, UWSpace.

Full publication list: [Google Scholar](https://scholar.google.com/citations?user=e-N_8owAAAAJ)

## Professional Interests & Activities

- Technical Writing: Maintain ML-focused blog at [www.jerpint.io]()
- Public Speaking: Regular invited speaker at AI conferences and workshops
- Hobbies: Rock climbing, hockey, guitar, drums, travel