WORK EXPERIENCE
AI Engineer
pipikwan pêhtâkwan 09/2025 - Present Edmonton, Alberta, Canada- Finetuning LLMs catered to multi-purpose tasks for Indigenous Peoples, raising BERTScore performance by 20%.
- Building inference pipelines for billion-parameter models to serve large user bases, reducing response times to around 3 ms.
- Collaborating with multi disciplinary teams to integrate and test AI-powered sub-products.
Software Engineer
Apurba Technologies Ltd 12/2024 - 09/2025 Dhaka, Bangladesh- Worked on the core ML team of a government-adopted national industry-grade Bengali OCR system
- Built scalable and optimized RESTful APIs, improving performance by 10%.
- Co-created an improved word-level model for OCR, recording an accuracy of >95%.
- Collaborated with stakeholders to implement, research, and complete project deliverables 2-3 weeks before the deadline.
Lecturer (Contractual)
United International University 06/2024 - 10/2024 Dhaka, Bangladesh- Taught core CS concepts to students.
- Took class tests and term exams to monitor students' performance and provide informative feedback.
- Worked with the Dept. of CSE to audit students' marks for BAETE/IEB accreditation.
- Courses conducted: Object-Oriented Programming Lab, Data Structures and Algorithms 1 Lab, and Database Management Systems Theory.
Undergraduate Assistant
United International University 09/2023 - 05/2024 Dhaka, Bangladesh- Managed to facilitate course content and materials with the lab faculty, checked and evaluated assessments, and judged lab projects.
- Created the very first course materials for the country's first-ever undergraduate degree in Data Science.
- Courses supervised: Programming for Data Science, and Object-Oriented Programming for Data Science.
PROFESSIONAL SKILLS
Hyperledger
Fabric
Docker
Git
GitHub
EDUCATION
B.Sc in Computer Science & Engineering (Major: Data Science)
United International University
(2020 - 2024)
CGPA: 3.77/4.00, Major CGPA: 4.00/4.00
Thesis Title: AVLoS: Audio-Visual Long Text Scene Summarization
Thesis Supervisor: Prof. Dr. Swakkhar Shatabda
PUBLICATIONS
Journals
S M Jishanul Islam, Sahid Hossain Mustakim, Musfirat Hossain, Mysun Mashira, Nur Islam Shourav, MD. Rayhan Ahmed, Salekul Islam, Swakkhar Shatabda, and A.K.M. Muzahidul Islam. 2024. An audio video-based multi-modal fusion approach for speech emotion recognition. Under 2nd review at Knowledge-Based Systems (Elsevier).
Conference and Workshops
Md. Mehedi Hasan, S M Jishanul Islam, AKM Shahariar Azad Rabby, and Fuad Rahman. 2025. Beyond Characters: Position-Aware Metric and a Lightweight LLM for Low-Resource Bangla Punctuation Restoration. IEEE BigData 2025. Industry Work
Sahid Hossain Mustakim, S M Jishanul Islam, Ummay Maria Muna, Montasir Chowdhury, Mohammed Jawwadul Islam, Sadia Ahmmed, Tashfia Sikder, Syed Tasdid Azam Dhrubo, and Swakkhar Shatabda. 2025. Watch, Listen, Understand, Mislead: Tri-modal Adversarial Attacks on Short Videos for Content Appropriateness Evaluation. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2025, pp. 2955-2964. DOI: Link
S M Jishanul Islam, Sahid Hossain Mustakim, Sadia Ahmmed, Md. Faiyaz Abdullah Sayeedi, Swapnil Khandoker, Syed Tasdid Azam Dhrubo, and Nahid Hossain. 2024. MIMIC: Multimodal Islamophobic Meme Identification and Classification. 3rd Muslims in ML Workshop - NeurIPS 2024. DOI: Link
Sadia Ahmmed, Taimur Rahman, S M Jishanul Islam, Al-Momen Reyad, Sonjoy Dey, James Anthony Purification, and Md. Dewan Farid. 2024. E-MedViTR - Enhanced Vision Transformers with Registers for Biomedical Image Classification. 6th International Conference on Electrical Engineering and Information & Communication Technology (ICEEICT 2024). DOI: Link
RESEARCH EXPERIENCE
An audio video-based multi-modal fusion approach for speech emotion recognition
Co-supervised by: Md. Rayhan Ahmed, Dr. Salekul Islam, Dr. Swakkhar Shatabda, and Dr. A.K.M Muzahidul Islam.
Under Review at Knowledge-Based Systems (Elsevier) | [codebase]
We present an approach to classify human emotions by fusing audio and visual inputs. Our approach sets new state-of-the-art results while keeping the architecture very simple. Additionally, we present a new frame filtering strategy to overcome the problem of spatiotemporal redundancy.
ChimeraBreak: a novel coordinated tri-modal attack on MLLMs
Supervised by: Dr. Swakkhar Shatabda.
Published at Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops. | [codebase] [arxiv] [CVF]
We introduce SVMA, an adversarial dataset for content moderation in short-form videos, and ChimeraBreak. This coor- dinated strategy exposes systemic safety flaws in leading MLLMs for content appropriateness evaluation. We experiment with open and closed source MLLMs and utilize LLM-as-a-Judge to evaluate ethical reasoning and confidence.
MIMIC: Multimodal Islamophobic Meme Identification and Classification
Supervised by: Nahid Hossain.
Accepted at MusIML Workshop - NeurIPS 2024. | [codebase] [arxiv]
Anti-Muslim hate speech has emerged within memes, characterized by context-dependent and rhetorical messages using text and images that seemingly mimic humor but convey Islamophobic sentiments. This work presents a novel dataset and proposes a classifier based on the Vision-and-Language Transformer (ViLT) specifically tailored to identify anti-Muslim hate within memes by integrating both visual and textual representations.
BhaShammo: IPA Transcription of Bengali Regional Dialect using Dialect Guided Tokens
Co-supervised by: Dr. Swakkhar Shatabda and Dr. Farig Sadeque.
In Progress. | [codebase] [arxiv]
We present an approach to transcribe regional Bengali text to IPA by introducing the Dialect Guided Tokens (DGT) technique on a new dataset spanning six districts of Bangladesh. We provide the model with information on the regional dialect of the input text before generating the IPA transcription. This is the first time this problem has been solved.
PROJECTS
NurtureAid
Developed a real-time AI-based cross-platform mobile application that simplifies care through caretakers. Led the app's backend and frontend development as a full-stack engineer.
Tech Stack: React Native, Node.js, Flask, MongoDB, Firebase, LLaMA-Index, PyTorch.
Acknowledgements: Champion of the UIU CSE Project Show, Software Engineering Laboratory; Champion of the Hult Prize OnCampus round in UIU; Selected for the Hult Prize Summit in Boston.
Quest Aid
Developed an AI-driven web application to manage the ECAs of students, clubs, universities, and organizations under the same platform. Led the app's backend, frontend, and AI development as a full-stack engineer.
Tech Stack: React.js, Spring Boot, Flask, MySQL, Langchain.
iamspecial.com
Developed a web app built to tackle the lack of misinformation, accessibility and guidance for people with special needs. Led the app's backend and frontend development as a full-stack engineer.
Tech Stack: HTML, CSS, JavaScript, PHP, MySQL.
Acknowledgements: Champion of the UIU CSE Project Show, Database Management Systems Laboratory.
ECMS-Desktop
Developed an ECA Management System for uniting the ECAs of students, clubs, universities, and organizations under the same platform. Led the app's backend and frontend development as a full-stack engineer.
Tech Stack: Java, JavaFX, MySQL.
Acknowledgements: Champion of the UIU CSE Project Show, Advanced Object-Oriented Programming Laboratory.
ACHIEVEMENTS
Best Performing Engineer, Project SD-12 awarded by Apurba Technologies Ltd. October 2025
Winner, Harvard Health Systems Innovation Lab Hackathon 2025 Dhaka Hub April 2025
Academic Excellence Scholarship 2020 — 2024
Champion, Bhashamul: Bengali Regional Text to IPA National NLP Datathon March 2024
Champion, Hult Prize OnCampus Round March 2024
Top 15 Finalist, Hult Prize Summit in Boston, USA June 2024
Finalist, ICT Innovation Grant (for NurtureAid) December 2023
Gold Award (Champion), International Blockchain Olympiad IBCOL 2023 November 2023
Champion, Software Engineering Lab September 2023
Silver Award (1st Runner Up), Bangladesh Blockchain Olympiad BCOLBD 2023 July 2023
1st Runner Up, Intra-University Deep Learning Sprint January 2023
Champion, Database Management Systems Lab May 2022
Champion, Advanced Object Oriented Programming Lab January 2022
The Daily Star Award 2017
Academia High Achievers Award 2017