Asif Al Faisal

IIT, University of Dhaka

Data Analyst

Currently working and worked on the following projects and programmes:

• Big Data Analytics for Climate-Smart Agriculture in South Asia (BigData2CSA) Project.
• Asia Regional Resilience to a Changing Climate (ARRCC) Research Programme.
• Tailor-Made Training Plus (TMT+) for Bangladesh – An Orange Knowledge Programme.

Engineer (Team Lead)

As a team lead my responsibilities were to, manage a team of four junior engineers and two network transmission engineers, analyze daily network traffic data along with power consumption data and generate statistical summary reports of power to performance ratio to aid in the decision making of upper managment.

Engineer

My responsibilities as an Engineer were to, monitor telecom network of two major mobile network operators (GP & Robi) in Bangladesh, provide countrywide remote network support on a roster basis and development of multiple python subroutine to automate multiple reports.

The goal of this research is to develop QSAR/QSPR models for predicting physicochemical properties and environmental fate endpoints of chemicals for regulatory purposes. The steps of this research are,

• Compare Graph Neural Net (GNN) model performance with an existing work where they have used weighted K-Nearest Neighbors (kNN).
• Train GNN models with multiple datasets to get more robust prediction of physicochemical properties.

In this research project,

• Primary goal is to collect and interpret a wide variety of primary agronomic management and socioeconomic data from tens of thousands of smallholder rice and wheat farmers.
• After data collection, I developed a novel multivariate method for outlier detection and imputation.
• Then, I developed R-scripts to extract secondary remote sensing and weather data in the geolocations of the survey and merged with primary data.
• I am also heavily involved model interpretation part where we used explainableAI (XAI) methods like LIME, ALE (manuscript writing ongoing).

In this research project,

• I developed a new crowd sourcing web-based tool that automatically harvests media reports on wheat rust disease occurrence and locations from the media in South Asia. These data then provide information to drive disease forecast models.
• Pilot testing of the tool has resulted in promising results, finding sources from where wheat rust diseases are appearing and spreading, significantly aiding meteorologically aided disease forecasting work.

This was my MS thesis work that went into publication in an IEEE conference. This was a Natural Language Processing (NLP) task where the goal was to develop a Convolutional Neural Net (CNN) for predicting personality traits (extroversion, agreeableness etc.) from written essays and compare how different activation functions influence prediction results.

Tailor-Made Training Plus (TMT+): Orange Knowledge Programme

This is a training programme in collaboration with the Bangladesh Agricultural Research Institute (BARI), the Faculty Geo-Information and Earth Observation Science (ITC) and CIMMYT. There are four stages of this programme where each stage consists of multiple courses. Among these courses, I got to work as a teaching assistant in following two courses.

• Basics of Python
• Algorithms
• Scientific Libraries
• Geocomputing

• Introduction to Geospatial data analysis
• Python Review
• Database
• Spatial Database
• Exploratory Data Analysis and Exploratory Spatial Data Analysis
• Introduction to Machine Learning
• Unsupervised Learning
• Decision Tree and Random Forest
• Artificial Neural Networks: Remote Sensing Image Classification