Pedram Farghadani-Chaharsooghi

Pedram Farghadani-Chaharsooghi

PhD in Operations Research

Professional Experience

Data Scientist

Yeji Data Lab, Montreal, QC
Sep 2025 – Present

  • Focusing on optimization using Operations Research.

Data Scientist

DataSphere Lab | McGill University, Montreal, QC
Jul 2025 – Sep 2025

  • Conducted exploratory data analysis on large-scale transactional sales data from Infasco, including cleaning, anomaly detection, and outlier handling.
  • Developed and deployed demand forecasting models using XGBoost and LightGBM, improving prediction accuracy for sales planning.
  • Applied advanced feature engineering (one-hot encoding, temporal feature extraction) to enhance model performance.
  • Optimized models through hyperparameter tuning and rigorous evaluation (RMSE, SMAPE), and visualized results to support business decision-making.

Operations Research Specialist

CIRRELT, Montreal, QC
Jan 2021 – Sep 2025

  • Built a robust Retrieval-Augmented Generation (RAG) system using LangChain, Azure OpenAI embeddings, and FAISS to retrieve context from disaster planning docs and answer technical user queries with an LLM, with fallback text search and query classification logic. ▶️ Demo: RAG for ACFs
  • Developed a Streamlit + LangChain AI assistant to support hospital evacuation planning during disasters. The app interprets free-text queries, extracts structured planning parameters using prompt-based logic and few-shot examples, and invokes a backend optimization engine. Integrated with Azure OpenAI and dynamic user feedback.
    ▶️ Demo: LLM for ACFs
  • Designed and implemented Reinforcement Learning models (Q-Learning, Deep Q-Learning) integrated with Adaptive Large Neighborhood Search (ALNS) to optimize decision-making processes under uncertainty in complex environments, including disaster management and resource allocation.
  • Developed a Stochastic Dual Dynamic Programming (SDDP) algorithm to solve multi-stage stochastic optimization models, improving resource allocation efficiency in dynamic settings such as disaster response.
  • Applied advanced scenario clustering methods (K-means++, Self-Organizing Maps) combined with stochastic optimization techniques to enhance forecasting, simulation, and decision-making processes in large-scale planning problems.
  • Designed and benchmarked Benders decomposition algorithms to identify optimal locations for temporary hospitals, evaluating their performance against the Progressive Hedging algorithm.

Teaching Assistant

Concordia University, Montreal, QC
Fall 2024 – Winter 2025

  • Served as Teaching Assistant for Scheduling Theory and Quantitative Methods in Healthcare Systems, under the supervision of Prof. Hossein Hashemi Doulabi, providing support to students in their coursework and projects.

Lab Demonstrator

Concordia University, Montreal, QC
Fall 2024

  • Assisted Prof. Gerard Gouw in delivering and managing labs for the course Human Factors Engineering, ensuring smooth execution of laboratory sessions with more than 30 students.

Software Trainer

FaraDars Educational Institution, Tehran, Iran
Summer 2019

  • Developed and designed an online course titled How to Balance a Production/Assembly Line using the Flexible Line Balancing (FLB) Software.
  • Facilitated the enrollment and successful completion of over 900 students in the course.