Open to opportunities

Aarooj Yashin

@aaroojyashin

Message

I am an ML researcher and developer specializing in LLM pipelines, forecasting, and MLOps.

Canada

Message

What I'm looking for

I seek an ML role focused on LLM pipelines, time-series forecasting, and production MLOps at a collaborative company that values research-to-production impact and stakeholder-facing solutions.

I am a post-graduate researcher and machine-learning developer who builds production-ready models and automated pipelines. I design async LLM-based document-extraction workflows and predictive pricing systems that deliver measurable business impact.

My work has reduced manual document validation time by 70% and produced a predictive pricing model with $500K in projected savings and a 40% increase in offer acceptance. I combine statistical methods, SHAP-based interpretability, and robust feature engineering to meet governance requirements.

I also develop proxy models and forecasting architectures that drastically cut compute: a proxymodel reduced CPMD simulation load by 1,000x. I value stakeholder collaboration, reproducible MLOps practices, and translating research into deployable solutions.

Experience

Work history, roles, and key accomplishments

Current

Associate Machine Learning Developer

Current

AltaML

Oct 2024 - Present (1 year 8 months)

Designed an async LLM-based document IDP pipeline using OpenAI APIs, reducing manual validation time by 70% and automating validated exports for downstream use. Built dynamic pricing models that reduced excessive discounts by 30%, increased offer acceptance by 40%, and delivered $500K in projected savings.

Python Prompt Engineering Pydantic AsyncIO XGBoost Random Forest GitHub

Graduate Research Assistant

University of Alberta

Sep 2023 - Sep 2025 (2 years)

Developed time-series forecasting models using Transformer encoders and LSTM layers to predict reaction trajectories and optimized training with adaptive rate scheduling. Created CNN-based embeddings and a proxy model that reduced CPMD simulation compute by 1000×, shrinking two months of calculations to 48 hours (results under review).

Python TensorFlow Transformers CNN Time Series Probabilistic Attention Optimization Adaptive Learning Rate

Education

Degrees, certifications, and relevant coursework

University of Alberta

Master of Science (Thesis), Chemical and Materials Engineering

2023 - 2025

Grade: 3.75/4.00

Thesis-based Master of Science in Chemical and Materials Engineering at the University of Alberta (2023–2025), GPA 3.75/4.00; research focused on predictive modeling and proxy simulations to accelerate CPMD calculations.