Kawshik Kumar Paul
Software Engineer | Researcher
B.Sc. Engg. (CSE), BUET
Software Engineer at REVE Systems
Research Interests
About
I am a Software Engineer looking to pursue a Master's or PhD in Artificial Intelligence and Software Engineering. I am mainly interested in Large Language Models (LLM), Natural Language Processing (NLP), Computer Vision, Speech AI (Text-to-Speech and Speech Recognition), Human-Computer Interaction (HCI), Computer Security, Software Engineering, and Distributed Systems. I have a strong academic background in Computer Science and Engineering, with hands-on experience in research, software development, and AI model finetuning. I am seeking opportunities to contribute to cutting-edge research and development in these areas.
Explore: Education · Work Experience · Research · Projects · Skills · Contributions · News · CV
Prior to this, I completed my BSc. (Engg.) in Computer Science and Engineering (CSE) from Bangladesh University of Engineering and Technology (BUET), where I gained foundational knowledge in software development and research methodologies.
My current research interests include:
- Large Language Models (LLM) & NLP: Model internals and configuration (architecture, internal parameters, and behavior), full and LoRA-based finetuning, and LLM-based applications — structured planning and validation, semantic data mapping, and agentic systems.
- Speech AI — Text-to-Speech (TTS) & Automatic Speech Recognition (ASR): Finetuning TTS and ASR models, dataset preparation, model quantization and Real-Time Factor (RTF) optimization, and low-resource adaptation techniques like LoRA.
- Computer Vision & Human-Computer Interaction (HCI): Emerging areas of interest alongside my speech and language work.
- Computer Security: Applied security problems spanning threat detection, malware analysis, and API/web security.
- Software Engineering & Distributed Systems: Scalable backend architecture, distributed state synchronization, and enterprise system design.
Alongside this, I'm working on a couple of longer-term efforts on the side: I'm writing a book on interview preparation for entry-level software engineering jobs in Bangladesh (still an ongoing, unpublished project), and I've recently started putting together a small team to build open-source software together for learning purposes. It's early days for the team, so more on that in time.
I am actively seeking graduate research opportunities and would welcome the chance to discuss how my background and interests might align with your research group. If any of the above resonates with your work, I'd love to connect and learn more about potential openings or collaborations under your supervision.
Feel free to reach out to discuss research, collaborations, or opportunities.
News
Submitted "TroBot: A Trojan Attack Using a Chatbot" to NSysS 2026 (13th International Conference on Next-generation Computing, Communication, Systems and Security)
Contributed the first LoRA (Low-Rank Adaptation) finetuning implementation to k2-fsa/OmniVoice, a large open-source TTS repository built on Xiaomi's OmniVoice model
Submitted "A Low-Cost ATmega32-Based Embedded System for Automated Patient Queue and Health Data Management in Private Medical Chambers" to NSysS 2026 (13th International Conference on Next-generation Computing, Communication, Systems and Security)
Formed and began supervising a small team building open-source software for learning purposes — a volunteer effort, still in its early stages
Submitted "The Choice Can Be the Attack: Auditing Aligned Backdoors in LLM Agents" to TACL (Transactions of the Association for Computational Linguistics)
Published a finetuned Bengali Text-to-Speech (TTS) model (CosyVoice3) on HuggingFace, adding Bengali support the base model didn't originally have, along with a public HuggingFace Space demo open to everyone
Acknowledged in "AutoTestGen: A Multi-Agent LLM Based Pipeline for End-to-End Web Test Generation from Natural Language Descriptions" for developing ground-truth test suites and assisting with evaluation and validation — an undergraduate thesis project supervised by Sukarna Barua at BUET CSE
Began writing a book on interview preparation for entry-level software engineering jobs in Bangladesh — ongoing
Recognized with a salary increment for outstanding performance
Joined REVE Systems as a Software Engineer
Received an offer for Software Engineer position at Samsung R&D Bangladesh
Began working as a Software Programmer on a project with Grameen Bank
Graduated with B.Sc. in Computer Science and Engineering (CSE) from BUET
Defended undergraduate thesis on a patient-centric blockchain framework for secure electronic health record management, ahead of graduating
Commenced undergraduate studies in Computer Science and Engineering at BUET
Got admitted to Bangladesh University of Engineering and Technology (BUET) in Computer Science and Engineering
Achieved 86th rank in BUET admission test - the most difficult engineering admission exam in Bangladesh with over 100,000 applicants competing for 1,100 available seats
Qualified for admission to a Public Medical College - the most difficult medical college admission in Bangladesh with over 100,000 applicants competing for only 3,200 available seats nationwide
Achieved qualifying rank for University of Dhaka - the top university for honors programs in Bangladesh with over 100,000 applicants competing for 1,500 available seats
Awarded Board Scholarship in Higher Secondary School Certificate (HSC) examination with Golden GPA 5.00
Awarded Board Scholarship in Secondary School Certificate (SSC) examination with Golden GPA 5.00
Completed Junior School Certificate (JSC) examination with Golden GPA 5.00
Got admitted to Rajuk Uttara Model School and College (RUMC), the best school in Bangladesh, for high school education
Received Board Scholarship in Primary Education Completion (PEC) examination, ranking 2nd in upazila with 93.5% marks
Ranked 5th in the admission test of Naogaon Govt. KD School, the best government school in Naogaon District
Ranked top 10 in Naogaon Government Zilla School admission test