Ayesheh Ahrari Khalaf

Ayesheh Ahrari Khalaf

Data Scientist Researcher
United Kingdom

About Me

Data Scientist Researcher with experience in AI-driven cognitive behavioral therapy systems, data analysis, technical operations, and research. Skilled in Python, scikit-learn, NumPy, pandas, TensorFlow, NLP, NLU, SQL, A…

Experience

Data Scientist Researcher

International Islamic University of Malaysia (IIUM)
Feb 2020 - Present · 6 years 8 months

• Modeling an AI-driven cognitive behavioral therapy system using Python (scikit-learn, NumPy, pandas, TensorFlow), deep learning (NLP, NLU), a Persona Perception bot, and SQL, partially computed in AWS.
• Achieved State-of-the-Art Accuracy by implementing a 1.5B parameter transformer model stored in Hadoop, outperforming 7 language modeling datasets and garnering positive human evaluations.
• Evaluated system on PERSONA-CHAT, surpassing state-of-the-art baselines with 82.2% accuracy on Hits@1(%) in original data and 68.8% on revised data, visualized using R.
• Collaborated seamlessly with colleagues to co-author and publish three impactful journal papers.

Data Scientist Researcher

International Islamic University of Malaysia (IIUM), Kuala Lumpur, MY
Feb 2020 - Jul 2023 · 3 years 5 months

<p>Modeled an AI-driven cognitive behavioral therapy system using Python, scikit-learn, NumPy, pandas, TensorFlow, deep learning, NLP, NLU, a Persona Perception bot, SQL, and AWS.<br>
Achieved state-of-the-art accuracy by implementing a 1.5B parameter transformer model stored in Hadoop.<br>
Outperformed 7 language modeling datasets and garnered positive human evaluations.<br>
Evaluated the system on PERSONA-CHAT, surpassing state-of-the-art baselines with 82.2% accuracy on Hits@1(%) in original data and 68.8% on revised data.<br>
Visualized results using R.<br>
Co-authored and published three journal papers with colleagues.</p>

Technical Operations Consultant

Dassault Systèmes, Quintiq, Petaling Jaya, MY
Nov 2017 - Jan 2020 · 2 years 2 months

<p>Assessed and resolved technical issues for over 300 customers across 80 countries with individually customized Quintiq applications.<br>
Achieved a 95% customer satisfaction rate while meeting Service Level Agreements (SLAs).<br>
Streamlined and automated the timesheet recording process.<br>
Reduced manual errors by 90%.<br>
Saved an average of 10 hours per week for the team.<br>
Operated Root Cause Analysis (RCA) investigations for emergency hot fixes.<br>
Reduced system downtime by 30%.<br>
Improved system stability.<br>
Resolved integration issues between middleware technologies and customized applications including SAP, PTV, and BIRT.<br>
Managed incidents related to VMs and databases including Oracle and SQL Server.</p>

Data Analyst and team leader

International Islamic University of Malaysia (IIUM), Kuala Lumpur, MY
Jan 2016 - Mar 2017 · 1 year 2 months

<p>Engineered ShifaBand integrating multiple sensors, a DIY Arduino microcontroller, Bluetooth, and an Android app.<br>
Conducted A/B testing.<br>
Implemented a cost-effective solution that was 50% more affordable than comparable alternatives.<br>
Headed a five-member team to propose, design, and implement the ShifaBand project.<br>
The project was recognized as the best project.<br>
Implemented a Cost Minimization Technique in Geo-Distributed Cloud Computing Data Centers using Python and Gurobi academic solver.<br>
Improved cost optimization efficiency by up to 20%.<br>
Applied ETL techniques and oversaw the data science pipeline.<br>
Performed cleansing, wrangling, visualization, modeling, and interpretation.<br>
Published two research papers.</p>

Malaysia’s representative at ACM-ICPC

International Islamic University of Malaysia (IIUM), Kuala Lumpur, MY
Jan 2014 - Jan 2016 · 2 years

<p>Champion in Aturkreatif Open Programming Competition @ USIM 2016.<br>
1st Runner-Up at 2016 ACM-ICPC Malaysia Al-Khawarizmi National Programming Contest.<br>
4th place at 2014 ACM-ICPC Malaysia Al-Khawarizmi National Programming Contest.<br>
Asia Regional Singapore site 2015 - Honorable Mention.<br>
Kuala Lumpur site 2014.</p>

Research Assistant | ELIVE II & LTE-Plus Projects

Telekom Malaysia Research and Development, Cyberjaya, MY
Jun 2015 - Aug 2015 · 2 months

<p>Achieved 97% accuracy in developing a real-time weather website.<br>
Integrated Waspmote kit and Raspberry Pi using CoAP protocol.<br>
Utilized APIs, Excel, and Git for improved data reliability.<br>
Led to a research paper publication.<br>
Implemented an HMM engine utilizing the EM Statistics algorithm.<br>
Incorporated a Two-component GMM model that improved accuracy in power disaggregation using C++ and big data analysis.<br>
Developed a website using HTML, CSS, PHP, JavaScript, and MySQL.<br>
Presented the three steps of the power disaggregation algorithm to administrators.<br>
Enhanced data visualization and user interaction.</p>

Skills

Data Evaluation Python SQL Database Academic Research Data Science Data Representation Hadoop NumPy TensorFlow Amazon Web Services (AWS) Natural Language Processing scikit-learn pandas TensorFlow NLP NLU SQL C++ HTML CSS PHP JavaScript MySQL Git Excel Gurobi Arduino Bluetooth Raspberry Pi CoAP APIs Middleware SAP PTV BIRT Oracle SQL Server ETL Data Wrangling Data Cleansing Machine Learning Deep Learning A/B Testing Root Cause Analysis Microcontroller
Report this Profile?