About Me
I am a Senior Software Engineer specializing in Machine Learning and Artificial Intelligence with extensive experience in designing and implementing AI-powered solutions. Currently working at IQVIA, I specialize in GenAI, Large Language Models (LLMs) and multi-agentic orchestration systems.
My expertise spans across fine-tuning and deploying Foundation Models, building AI assistants and leading cross-functional teams to deliver innovative products. I have a strong background in machine learning, deep learning, and cloud platforms.
With a Master's degree in Computer Science & Engineering from Rajshahi University of Engineering & Technology, I combine academic excellence with practical industry experience to solve complex business challenges through AI and machine learning.
Software Skills
Machine Learning Libraries
Data Analytics Tools
Programming Languages
Databases
Web Frameworks
Platforms
Professional Experience
Software Development Engineer 3 | Sr. Machine Learning Engineer
IQVIA, Dhaka
April 2024 - Present
- Designed & improved IQVIA Agentic AI Assistant
- Implemented multi-agentic orchestration
- Fine-tuned & deployed Foundation Models (Llama 2, Llama 3)
- Led communication for product requirements with clients
- Managed collaboration for successful product delivery
Software Engineer-2 | ML Engineer
BRAIN STATION 23 PLC., Dhaka
January 2024 - April 2024
- Managed projects with cross-functional teams and clients to design, implement, and deploy features
- Practiced Scrum methodologies throughout the project lifecycle
- Designed Proof of Concepts (POCs), ML Workflows to address specific business challenges
- Led ML R&D with a focus on generative AI
Software Engineer-1 | ML Engineer
BRAIN STATION 23 PLC., Dhaka
July 2023 - December 2023
- Designed Proof of Concepts (POCs), ML Workflows to address specific business challenges
- Conducted ML R&D with a focus on generative AI
- Leveraged Large Language Models (LLMs) to tackle natural language tasks while reducing cost & response time
- Implemented Graph Database to represent complex relations between entities for better insight
Associate Software Engineer-2 | ML Engineer
BRAIN STATION 23 PLC., Dhaka
October 2022 - June 2023
- Conduct ML R&D with a focus on generative AI
- Experienced in POC design, ML workflows and pipelines
- Experienced in creating APIs for ML models
Associate Software Engineer-1 | ML Developer
BRAIN STATION 23 PLC., Dhaka
March 2022 - September 2022
- Conducted analysis on various datasets and implemented machine learning models
- Developed Generative Adversarial Networks (GANs) for synthetic data generation
- Created web-based machine learning solutions using React and NodeJS
- Containerized applications and deployed them on the cloud platforms
- Obtained Alteryx Certification
Software Engineer Trainee
R&D Lab, BRAIN STATION 23 PLC., Dhaka
November 2021 - February 2022
- Utilized machine learning and deep learning algorithms with a focus on computer vision
- Conducted data preprocessing and cleaning tasks
- Built predictive models using machine learning algorithms
- Analyzed and interpreted complex data sets
- Fine-tuned pre-trained models on collected datasets to address specific tasks
- Gathered and analyzed data, as well as visualized different datasets
Education
Master of Science (Engg.)
Dept. of Computer Science & Engineering
Rajshahi University of Engineering & Technology
RAJSHAHI-6204, Bangladesh
CGPA: 3.92/4.00
Thesis remaining
Bachelor of Science (Engg.)
Dept. of Computer Science & Engineering
Rajshahi University of Engineering & Technology
RAJSHAHI-6204, Bangladesh
CGPA: 3.75/4.00
First Class with Honors (5th)
Thesis
Generative Adversarial Network based Synthetic Data Generation System
Period: April 2021 - November 2022
Supervisor: Barshon Sen, Assistant Professor, Dept. of Computer Science & Engineering, RUET
Conducted research focused on enhancing synthetic tabular data generation using Conditional Tabular Generative Adversarial Networks (CTGAN). Performed extensive exploratory data analysis (EDA), addressing issues such as missing values, skewed feature distributions, class imbalance etc. through targeted preprocessing techniques. Developed and applied a robust evaluation framework, including principal component analysis (PCA), statistical tests (Kolmogorov-Smirnov, logistic detection etc.), mutual information analysis, feature importance analysis etc. to validate synthetic data quality. Achieved high validation accuracy using machine learning algorithms, notably attaining 98.2% and 96.3% accuracy with XGBoost on real and synthetic data, respectively.
Projects
LEADRS Probable Case Narrative Generation and Review Tool
Technical Responsibility:
As a Machine Learning Engineer, I spearheaded the development of an innovative feature for the LEADRS project, focusing on the generation and review of probable cause narratives using LLMs. My role involved integrating this feature with the existing LEADRS system through Flask APIs, leveraging LLMs for generating contextually accurate and legally compliant narratives. Additionally, I created a review tool for ensuring narrative accuracy and relevance, with a keen focus on legal and ethical data handling.
Description:
In the LEADRS project, my key contribution was enhancing the system's DUI/DWI case handling efficiency for Texas law enforcement. I developed an AI-driven tool for generating and reviewing probable cause narratives, which significantly eased the report writing workload. This tool, empowered by LLM technology, produces precise, legally sound narratives tailored to specific cases, marking a notable advancement in AI's application in legal reporting and law enforcement.
Maruboshi OM Keyword Generation
Technical Responsibility:
As an ML Engineer, I developed Flask APIs for Maruboshi OM, enabling multilingual keyword generation to optimize car companies' SEO. Using BeautifulSoup and OpenAI LLM Models, I efficiently extracted meaningful content from each page, enhancing keyword extraction capabilities. Efficient data processing was ensured through batch processing, file handling, language detection, and Flask-CORS integration for secure data transmission across platforms.
Description:
The Maruboshi OM project is a keyword generation service for OM manuals in 21 different languages. Each OM manual consists of over 1600 pages, and our AI system efficiently generates keywords for each page to enhance car companies' SEO efforts. With our system, complete OM manuals can be processed within 2 minutes, providing quick and valuable insights for search engine optimization.
MediLynq
Technical Responsibility:
In this project, I was involved in managing Holter devices, which are typically delivered to hospitals and patients. Key responsibilities included analyzing Holter data (in EDF format) using AI algorithms to generate insightful ECG reports.
Description:
MediLynq is a comprehensive Holter device management project that streamlines the delivery of these devices to healthcare institutions and patients. A crucial aspect of this project was analyzing the received Holter data (in EDF format), for which we employed advanced AI algorithms. The outcome of this analysis was translated into detailed and insightful ECG reports, providing critical health information to medical professionals and contributing to better patient care.
DIGITAL INFLUENCER
Technical Responsibility:
As an ML Engineer in this project, my tasks involved using Natural Language Processing and Generative AI techniques, particularly the GPT-2 model, to create a system capable of generating human-like tweet posts. These posts included mentions, hashtags, numerical facts, and links pertinent to recent events.
Description:
The Digital Influencer project harnessed the capabilities of Natural Language Processing (NLP), Generative AI, and GPT-2 to simulate human-like interactions on Twitter. It was designed to generate tweets reflecting current events, complete with relevant mentions, hashtags, numerical facts, emoticons and links. This project essentially aimed to automate the creation of engaging, topical, and contextually relevant social media content, enabling greater reach and interaction in the digital space.
E-KYC (ELECTRONIC KNOW YOUR CUSTOMER)
Technical Responsibility:
As an ML Engineer, my responsibilities included building REST APIs via Flask to serve deep learning model services (including Face Recognition, Verification, and NID Parsing OCR) to the client application. My role also involved improving the performance of these deep learning models.
Description:
The e-KYC project, an online onboarding system for AB Bank, aimed to simplify and digitize the process of account creation. It offers a paperless registration process where users can verify their identity through automated procedures such as face verification and National Identification (NID) information verification.
Technical Responsibility:
In the capacity of a Data Analyst, I executed Descriptive, Diagnostic, Predictive, and Prescriptive analytics on a dataset comprising 550K records in the retail sector.
Description:
This project entailed offering Data Analysis as a Service (DAAS) for the retail industry. The work revolved around conducting various levels of analysis - Descriptive, Diagnostic, Predictive, and Prescriptive - on a retail industry dataset with over 550,000 records. The objective was to identify actionable insights to enhance efficiency and profitability in the retail sector.
SENSE-23
Technical Responsibility:
As an ML Engineer, I worked on a comprehensive solution for product identification that included the development of an Image Generating Tool, an Image Annotation Tool, and a Machine Learning Object Recognition model.
Description:
The Sense-23 project served as a full-stack solution for product identification, enabling automation and precision in object recognition. My contributions as an ML Engineer included the development of an Image Generating Tool, Image Annotation Tool, and an Object Recognition model based on Machine Learning, harnessing techniques from Deep Learning and Computer Vision. This suite of tools and applications significantly enhanced product identification processes and accuracy.
Certifications
Publication
Location based Probabilistic Load Forecasting of EV Charging Sites: Deep Transfer Learning with Multi-Quantile Temporal Convolutional Network
Conference: International Conference on Architecture of Computing Systems
Date: April 2025
Get In Touch
Contact Information
Connect With Me
© 2025 Md. Aukerul Moin Shuvo. All rights reserved.