Mohammad Karimi | Senior AI Engineer

Hi, I'm Mohammad Karimi.

A(n)

Senior AI Engineer focused on reliable, high-performance LLM systems and applied machine learning that moves real products.

About

I’m a Senior AI Engineer at Proofpoint with 8+ years of experience building end-to-end ML systems — from model development to production deployment. I focus on modern NLP/LLM systems (high-throughput inference, cost/performance optimization, and evaluation) and enjoy solving hard problems with pragmatic engineering.

Languages: Python, SQL, C++, JavaScript, MATLAB, PHP, HTML/CSS
NLP: NLTK, Spacy, Gensim, Hugging Face, Stanza
Databases: MySQL, Microsoft SQL Server, SQLite, PostgreSQL, Pandas
Libraries: Tensorflow, Pytorch, Keras, Sklearn, Trax, OpenCV, YOLO, Numpy
Tools: Azure ML Studio, Linux, Bash, Git, Docker, Google Earth Engine, Google Street View, ROS
Soft Skills: Teamwork, Leadership, Communication, Work Ethic, Time Management, Creativity

I am seeking a challenging position that will enable me to combine my skills in Data Science and ML Development, while fostering professional growth, engaging experiences, and personal development.

Experience

Proofpoint

Senior AI Engineer

Architected and deployed a high-performance Redaction API on Kubernetes using a hybrid of BERT transformers and rule-based patterns, delivering a 122× cost reduction vs. AWS Guardrails while maintaining high precision
Deployed 20+ open-weight models on Kubernetes using vLLM for classification and reasoning; enabled KV caching and CPU offloading to achieve 10× lower inference costs than AWS Bedrock
Optimized AWS Nova models on Bedrock for classification by integrating tool use (function calling) and GEPA, improving overall dataset F1 by +40%
Developed and fine-tuned Longformer architectures for document classification, achieving 90%+ F1 and enabling Integrated Gradients for model explainability
Engineered lightweight classification pipelines using SentenceBERT, Model2Vec, and MLP classifiers, achieving an 82% F1-score on the holdout set for resource-constrained environments
Tools: Kubernetes, AWS, Bedrock, vLLM, Transformers, BERT, Longformer, SentenceBERT, Python

Jul. 2025 - Present | Toronto, ON, Canada

KPMG Lighthouse

Senior Data Scientist

Developed an AI-driven SOC report generator using NLP, Transformers, and local LLMs with instruction fine-tuning and inference acceleration, achieving 92%+ accuracy in information retrieval on GCP
Built and orchestrated 9 data science pipelines for Bell on Vertex AI (GCP), spanning data ingestion to model deployment using Cloud Composer, Pub/Sub, BigQuery, and more
Designed an end-to-end ML pipeline in Python as an accelerator for KPMG Lighthouse on GCP
Tools: Python, GCP, Vertex AI, BigQuery, Pub/Sub, Cloud Composer, Hugging Face, Transformers, Git

Jun. 2023 - Jul. 2025 | Toronto, ON, Canada

AltaML

Associate Machine Learning Developer

Developed and deployed a bias and hate speech classifier using OpenAI GPT-3.5 Turbo on Azure to classify hate speech and multiple bias categories from social media posts
Implemented and deployed a quick, scalable ML pipeline on Azure for the Government of Alberta to predict over 2.6B geospatial satellite imagery data points with 88%+ accuracy (cross-validation)
Tools: Python, Azure, Azure ML Studio, Scikit-learn, Git

Sep. 2022 - May 2023 | Edmonton, AB, Canada

Computer Science Dept.

Data Scientist and NLP Researcher

Developed and fine-tuned language models (n-grams and recurrent neural networks) to identify grammatically incorrect writing from big data resources with an accuracy of 97%+
Built a data + ML pipeline to predict Airbnb listing prices using XGBoost with a negligible error rate
Tools: Python, NLP, XGBoost, Tensorflow, Keras

Jan. 2021 - Aug. 2022 | Edmonton, AB, Canada

Iranian National Robotics Team

Software Engineer

Implemented a data acquisition pipeline to collect imagery datasets and trained Faster R-CNN for object recognition
Tools: Python, C++, ROS, Faster R-CNN, Linux, Bash

Sep. 2018 - Jan. 2020 | Tehran, Iran