Soroush Taheri

Data Scientist & ML Engineer

Experienced Data Scientist with 7+ years of expertise in Deep Learning, Computer Vision, and NLP. Proven track record of delivering measurable business impact through innovative ML solutions and published research in high-impact journals.

3 Publications
6+ Years Experience
29+ Successful Projects
Get In Touch View Achievements
Soroush Taheri

Core Expertise

Machine Learning Deep Learning Computer Vision NLP Time Series MLOps

About Me

I am a passionate Data Scientist and ML Engineer with a strong academic background and extensive industry experience. Currently working as an ML & Data Science Specialist, focusing on sales analytics and business optimization.

My expertise spans across machine learning, deep learning, and data science applications. I have developed advanced LSTM models for predictive analytics, implemented BERT-based embeddings for NLP tasks, and engineered computer vision systems for various applications. My work has resulted in significant improvements including 25% RMSE improvement in forecasting and 18% better accuracy in topic relationship detection.

With over 6 years of experience in both academic research and industry applications, I have worked on diverse projects ranging from energy forecasting and renewable integration to computer vision systems for urban analytics. My research has been published in high-impact journals including Scientometrics and presented at international conferences.

M.S. Computer Science: Software Engineering

Shahid Beheshti University
GPA: 3.9

ML & Data Science Specialist

Ostrich Industry Journal
Current Position

Industry Experience

Machine Learning & Data Science
Multiple Companies

Publications

A Comprehensive Review of Machine Learning Approaches for Decarbonisation and Energy Efficiency in the UK Building Sector

2025
Taheri, S., Ahmadi, E. and Izadpanah, A.
CIBSE IBPSA-England Technical Symposium 2025
DOI

An embedding approach for analyzing the evolution of research topics with a case study on computer science subdomains

2023
Harikandeh, S.R.T., Aliakbary, S. and Taheri, S.
Scientometrics, 128(3), pp.1567-1582
DOI

Research trend prediction in computer science publications: a deep neural network approach

2022
Taheri, S. and Aliakbary, S.
Scientometrics, 127(2), pp.849-869
DOI

Towards Study of Research Topics Evolution in Artificial Intelligence based on Topic Embedding

2021
Harikandeh, S.R.T., Aliakbary, S. and Taheri, S.
11th International Conference on Computer Engineering and Knowledge IEEE, pp. 406-411
DOI

Professional & Research Experience

ML & Software Specialist – Sales Analytics

Sep 2024 -- Present

Ostrich Industry Journal (Full-Time)

Tehran, Iran

  • ML & Business Optimization: Built end-to-end ML pipelines, trained predictive models, and developed real-time dashboards for sales analytics and executive reporting.

Data Scientist and ML Engineer

July 2022 -- June 2025

Energiser LTD (Part-Time)

Remote, London, UK

  • Energy Forecasting: Engineered Neural Networks for energy demand forecasting, achieving 7% cost reduction.
  • Renewable Integration: Implemented ensemble learning methods for power flow predictions, achieving 15% accuracy improvement.
  • Cloud Infrastructure: Architected scalable ML pipelines on AWS with MLOps practices, achieving 99.8% uptime.

Data Science Specialist

Oct 2020 -- Dec 2024

Freelance Professional (Part-Time)

Tehran, Iran

  • Computer Vision: Developed deep learning-based security systems and 3D image analysis pipelines using CNNs and transfer learning.
  • NLP: Created advanced NLP pipelines using BERT, GPT, and T5 models for sentiment analysis and automated report generation.
  • Urban Analytics: Engineered CNN-powered computer vision systems for cityscape analysis and urban planning applications.
  • Industrial AI: Designed ML systems for process optimization and predictive maintenance in industrial settings.

Data Scientist and Research Assistant

Dec 2018 -- April 2023

Data-Oriented Research for Software Analytics (DORSA) Laboratory

Tehran, Iran

  • Advanced LSTM Models: Engineered LSTM models for predictive analytics in scientific literature, achieving 25% RMSE improvement.
  • NLP & BERT-based Embeddings: Implemented BERT-based embeddings and topic modeling, improving topic detection accuracy by 18%.
  • Cross-domain Knowledge Discovery: Developed ML methodologies for bibliometrics, achieving 22% better prediction stability.
  • Research & Publications: Conducted empirical studies on scientometrics, contributing to peer-reviewed publications in high-impact journals.

Key Achievements & Impact

25% RMSE Improvement

Advanced LSTM models for predictive analytics in scientific literature

Deep Learning Time Series Research

7% Cost Reduction

Energy demand forecasting system that reduced operational costs for Energiser LTD

Energy Sector Cost Optimization Neural Networks

99.8% System Uptime

Production-grade ML pipelines on AWS with MLOps practices

MLOps AWS Production

18% Accuracy Boost

NLP techniques and BERT-based embeddings for topic relationship detection

NLP BERT Topic Modeling

4 Peer-Reviewed Publications

Published research in high-impact journals including Scientometrics

Research Publications Innovation

International Collaborations

Worked with researchers from Imperial College London, UTS, and Amazon

Collaboration Global Leadership

Education

M.S. in Computer Science

GPA: 3.9

Shahid Beheshti University

Tehran, Iran

Sep 2018 -- Jan 2021

Relevant Coursework:

Data Mining, Complex Networks, Cloud Computing, Advanced Database Systems, Distributed Systems, Advanced Software Engineering, Performance Evaluation

B.S. in Computer Science

GPA: 3.6

Islamic Azad University Central Tehran Branch

Tehran, Iran

Sep 2013 -- Dec 2017

Relevant Coursework:

Artificial Intelligence, Database Systems, Information Retrieval, Software Engineering, Operating Systems

Certifications

Google Data Analytics Specialization

Google

Issued: Feb 2025 Credential ID: US2GAV8Y6EU1
Show Credential

Neural Networks and Deep Learning

Coursera

Issued: Apr 2021 Credential ID: Q6VCPZVTBBQ2
Show Credential

Data Science Math Skills

Duke University

Issued: Aug 2025 Credential ID: IRXIDBW2H7RT
Show Credential

Mastering Key Performance Indicators (KPIs)

365 Data Science

Issued: Jul 2024 Credential ID: CC-113AE73DD9
Show Credential

Unsupervised Learning in Python

DataCamp

Issued: May 2021 Credential ID: 18973951
Show Credential

TOEFL iBT - Score: 110

ETS

Issued: Apr 2024 Credential ID: 4976304249799154
Show Credential

Get In Touch

Let's Connect

I'm always interested in discussing new opportunities, research collaborations, and innovative projects in data science and machine learning.

Google Scholar

View Profile