Mehdi Assefi, Ph.D.

# ad1q6p@r.postjobfree.com ï Mehdi Assefi +1-406-***-****

Work Authorization: US Green Card

Education

] Ph.D., University of Georgia Computer Science.

] M.Sc., Montana State University Computer Science.

] M.Sc., Azad University Software Engineering.

] B.Sc., Azad University Software Engineering.

Employment History

Aug 2020 - present ] Senior Data Scientist PNC Bank,

• Leadership

Model Risk Management: Identify, assess, and mitigate risks associated with NLP models.

Conversational AI Design: Lead the design and development of Conver- sational AI solutions across PNC.

• Training

Center of Excellence (COE): Provide training to AI teams on the design and implementation of AI solutions.

ML Solutions: Train the cross-functional teams on designing and de- veloping effective ML solutions for business problems using ML, Deep Learning, Generative AI, Transformers, and LLMs.

• Technical

Center of Excellence (COE): Apply technical skills in collaboration with AI teams to design and implement AI solutions. Employ machine learn- ing, deep learning, LLM, Generative Models, and NLP models to develop technically robust solutions for a variety of business problems from text processing and information extraction to predictive modeling and senti- ment analysis.

Conversational AI Design: Leverage technical knowledge to lead the de- sign and development of advanced Conversational AI solutions. Jul 2018 - Sep 2018 ] Data Science Research Intern - full time Fujitsu Labs, Sunnyvale (CA) Research area: NLP, Machine Learning - API classification and recommenda- tion system.

Dec 2017 - Apr 2018 ] Data Science Research Intern - full time NEC Labs, Cupertino (CA) Research area: Timeseries Forecasting and demand estimation on large data sets.

May 2013 - Aug 2013 ] Software Engineering Intern - full time Citrix, Goleta (CA) Relevant tasks: Remote Speech Recognition, experimental design, statistical analysis, A/B testing, ANOVA test, and causal inference. May 2020 - Aug 2020 ] Adjunct Professor University of Georgia 2004 - 2012 ] Faculty of Computer - Head of CS Department Azad University - Iran Selected Publications

US Patents

1 M. Assefi, M. Bahrami, and W. Chen, “Metadata-based api attribute extraction,” 2020. 2 M. Assefi, A. Hooshmand, and R. Sharma, “Deep learning approach for battery aging model,” 2019. Selected Conference/Journal Papers

1 M. Assefi, M. Bahrami, S. Arora, et al., “An intelligent data-centric web crawler service for api corpus construction at scale,” in International Conference on Web Services (ICWS), IEEE, 2022, pp. 385–390. 2 M. Assefi, A. Hooshmand, H. Hosseini, and R. Sharma, “Battery degradation temporal modeling using lstm networks,” in ICMLA’18, IEEE, 2018, pp. 853–858. 3 M. Assefi, E. Behravesh, G. Liu, and A. P. Tafti, “Big data machine learning using apache spark mllib,” in BIGDATA’17, IEEE, 2017, pp. 3492–3498.

4 A. P. Tafti, E. Behravesh, M. Assefi, et al., “Bignn: An open-source big data toolkit focused on biomedical sentence classification,” in BIGDATA’17, IEEE, 2017, pp. 3888–3896. 5 M. Assefi, G. Liu, M. P. Wittie, and C. Izurieta, “Measuring the impact of network performance on cloud-based speech recognition,” 2016, p. 19.

6 A. P. Tafti, A. Baghaie, M. Assefi, H. R. Arabnia, Z. Yu, and P. Peissig, “Ocr as a service: An experimental evaluation of google docs ocr, tesseract, abbyy finereader, and transym,” in Advances in Visual Computing: 12th International Symposium, ISVC 2016, Las Vegas, NV, USA, December 12-14, 2016, Proceedings, Part I 12, Springer, 2016, pp. 735–746. 7 M. Assefi, G. Liu, M. P. Wittie, and C. Izurieta, “An experimental evaluation of apple siri and google speech recognition,” vol. 118, 2015.

8 M. Assefi, M. Wittie, and A. Knight, “Impact of network performance on cloud speech recognition,” in 2015 24th International Conference on Computer Communication and Networks (ICCCN), IEEE, 2015, pp. 1–6.

Google Scholar

Awards and Recognition

2017 ] Grimes Fellowship

Franklin College of Arts & Sciences - University of Georgia 2016 ] Outstanding Researcher - Honorable Mention

Grimes School of Computing - Montana State University 2015 ] Best Paper Finalist Award

SEDE’15 Conference

Skills

Coding and DB ] Python, C++, Java, R, SQL, Postgresql ML skills ] Deep Learning(CNN, RNN, LSTM), Large Language Models(LLMs), NLU, Transfer Learning

General Skills ] Agile development, Linux, Databricks, Docker, AWS, Hadoop(HDFS), Spark, Hive, TensorFlow, Keras, Pandas, NumPy, Scikit-learn, Matplotlib, Seaborn, NLTK, Spacy, Scikit-learn, Pytorch, Huggingface

Selected Projects

] Information Extraction from Large Text Corpus Adverse vaccine reaction analysis using NLP

• Developed an intelligent big data analytics framework to analyze and visualize the content of scientific articles in identifying adverse vaccine reactions.

• The framework is developed on big data analytics solutions composed of Apache Spark and Elasticsearch No-SQL, CNN, LSTM, along with ML/NLP libraries like NLTK, NumPy, etc.

• The framework was used to analyze a dataset of over 1 million scientific articles on vaccines, resulting in the identification of over 10,000 new adverse vaccine reactions. Impact: The framework can be used by researchers and public health officials to better understand the risks and benefits of vaccines. The framework can also be used to develop new vaccines and improve vaccine safety.

] ML based web-crawler to construct API Corpus

• Developed a web-crawler service that constructs an API Corpus.

• The service is scalable and can collect a sheer number of APIs, intelligently filtering out non- related REST APIs using ML/NLP, and organizing pages per API title/provider.

• The API Corpus can be used for a variety of purposes, such as constructing Open API specifi- cations, accelerating digital transformation, and code generation. Impact: A novel web-crawler service was developed to construct a large corpus of API documen- tations. The service is scalable, intelligent, and organizes collected API documentations per API provider.

] TimeSeries Forecasting

• ATMs were frequently running out of cash, which was frustrating for customers and costly for the bank.

• We were spending a significant amount of money on transporting cash to and from ATMs.

• Our ATM cash flow forecasting was inaccurate, which led to overstocking and under stocking of ATMs.

Impact: This is an undergoing project. The potential saving from optimizing ATM cash flow can be significant. For example, a 2018 study by the ATM Industry Association found that US banks could save an estimated $1.7B billion per year by optimizing ATM cash flow[Link ].

] Classification and Recommendation on API Documentation Extracting API features from het- erogeneous API information and mapping them to OAS compliant formats:

• Developed and trained machine learning and NLP models to extract API features from hetero- geneous API information and map them to OAS compliant formats.

• The models were implemented in real-world scenarios to generate thousands of OAS-based APIs, whichwerethenvalidatedanddeployedintoacloud-based visual programming platform.

] Other Projects

Research projects on OCR, Fraud Deection, Timeseries forecasting, and clustering in collaboration with CDC, Marshfield Clinic, and University of Georgia.

Contact this candidate