Sign in

Data Scientist

Central, South Carolina, United States
May 16, 2019

Contact this candidate



864-***-**** Profile: LinkedIn, Webpage


Data scientist using mathematical modeling, optimization techniques, machine learning, and simulation to derive solutions and managerial insights for complex systems using data


Clemson University, Clemson, South Carolina, USA Aug 2013 - May 2019

Research Assistant

Proposing, designing, development and testing of new and efficient data-driven dynamic resource allocation models and policies

1.Real-time ambulance repositioning: Achieving 18% reduction in response time considering the proposed policies.

-Formulating a new design for emergency vehicle management considering unconventional decisions such as delaying low priority calls, redeploying vehicles to different bases, or relocating idle vehicles to cover for busy vehicles.

-The new design is based on a stochastic process, predicts future state of the system and reacts to future demand. It is calibrated and validated by real historical data, and it is run in real time.

-The process is formulated in dynamic programming framework. The real-time performance is made possible by deploying a multiple linear regression model to approximate the solution. The simulation is coded in VB.NET and uses CPLEX optimizer to estimate lower bounds.

2.Health care operations management: Designing syllabus and teaching a class

-Topics: Optimal facility location, blood inventory management, hospital job and personnel scheduling and queueing theories.

-I also taught the class how to code optimization problems utilizing Gurobi (gurobipy) in Python.

3.Optimal stopping of an online learning algorithm: New robust stopping rules ensuring a minimum level of accuracy.

-Proposed a new framework in sampling vs. stopping with respect to Net Present Value (NPV). Unlike current learning methods that only produce accurate rules when the unknown function is not flat, my proposal achieves a high probability of correct decision for any function.

-The new design uses a reinforcement learning algorithm and a diffusion approximation to evaluate stopping decision. The algorithms are coded in R and parallelized to reduce computation.

4.Optimal learning of unknown functions: Proposing a new and efficient learning technique for dose-response estimation.

-The new design reduced the complexity and significantly improved the computation effort required to derive new estimate of the unknown function using observational data.

-Currently, I am working on extending this design to allow for unknown correlation structures.

5.Cost-effectiveness analysis of diabetes interventions: Demonstrating that lifestyle interventions in patients with pre-diabetes symptoms reduces the risk of progression to diabetes, increases survival rate, and has financial benefits for the population.

-Simulating the progression of diabetes in U.S. using a coupled agent-based and population-based simulation model validated by real data. The coupled simulation model was a novel model coded in C++ and used regression, Cox hazard models and real data to estimate the parameters. The result confirmed that some lifestyle intervention in pre-diabetes stage not only can increase survival rates but may also reduce health care costs in the long run.

6.Predicting the outcome of state elections 2016: Developing a machine learning algorithm using Bayesian random walks and logit regression to predict the outcome of 2016 U.S. presidential election in each state.

Sharif University of Technology, Tehran, Iran Nov 2012 - Jun 2013

BSc Engineer

Responsible for implementing an analytical hierarchical process to select and evaluate candidates for management positions based on a numerical scoring system designed to reflect experience, knowledge, leadership qualities and colleagues feedback.

1.Human resource allocation in a university campus: Developing a linear programming method coupled with analytical hierarchical process to match each position with the best candidate.


Modeling: Dynamic Modeling, Stochastic Optimization, Machine Learning, Optimal Learning, Bayesian Inference, Simulation, Predictive Analysis, Forecasting, Graph Theory and Network Modeling

Programming: C++, R, Python, VB.NET, MATLAB, MySQL

Optimization: CPLEX, Gurobi, AMPL


Clemson University, Clemson, South Carolina, USA Aug 2013 – May 2019

PhD. Industrial Engineering

Sharif University of Technology, Tehran, Iran Sep 2008 – June 2013

B.S. Industrial Engineering


Nasrollahzadeh, A., Khademi, A., Mayorga, M., 2018, “Real-Time Ambulance Dispatching and Relocation," Manufacturing & Service Operations Management, 20(3) 467-480.

Nasrollahzadeh, A., Khademi, A., “Optimal Stopping of Adaptive Dose-Finding Trials," Submitted to Manufacturing & Service Operations Management

Khademi A., Shi, L., Nasrollahzadeh, A., Narayanan, H., Appel, L.J., Chen, L., “Compare Pre-diabetes Life-Style Interventions: An Integrated Microsimulation and Population Simulation Model," Submitted to Population Health Management

Nasrollahzadeh, A., Khademi, A., “Dynamic Programming to Response-Adaptive Dose-Finding Clinical Trials," submitted to INFORMS Journal on Computing


Best Graduate Student Paper Award 2018, Industrial Engineering Department, Clemson University

Contact this candidate