Biography & trajectory

Khan Raiyan Ibne Reza.

Computer Science researcher at North South University — Research Assistant (1 year), Undergraduate Teaching Assistant (3 semesters), author of 5 benchmark preprints.

The Research-to-Systems Loop

My academic work is driven by a central realization: meaningful AI research in low-resource linguistic contexts cannot rely on unmeasured assumptions; it requires rigorous empirical benchmarks and must be tested in working production software.

Rather than assuming multilingual foundation models generalize to South Asian low-resource languages, I design exhaustive diagnostic benchmarking protocols—such as KrishokChat (85k instances across 13 institutions) and Where Does Retrieval Fail? (1,000 queries over 2,882 knowledge nodes)—to measure exact failure boundaries in colloquial dialectal retrieval, domain-specific grounding, and multimodal document reasoning.

To complete the loop, I translate these diagnostic insights into 7 verifiable, deployed software systems (Next.js, FastAPI, PostGIS, MapLibre GL, CrewAI), testing resilient architectures under real-world conditions in Bangladesh.

Khan Raiyan Ibne Reza at North South University campus
NSU Campus · DhakaKhan Raiyan Ibne Reza

Khan Raiyan Ibne Reza

BSc in Computer Science & Engineering · NSU Dhaka

Bengali NLPMultimodal AI5 Preprints

Academic Appointments & Teaching Service

Research Assistant (RA)1-year appointment
North South University, Dhaka

Conducted empirical research on low-resource Bengali NLP, multi-stage RAG failure diagnostics, domain-specific safety refusal auditing, and dataset curation across 13 national agricultural institutions. Led authoring of 5 research benchmark preprints.

Undergraduate Teaching Assistant (TA)3 consecutive semesters
North South University, Dept. of History & Philosophy (DHP)

Mentored 200+ undergraduate students in critical argumentation, scientific methodology, and analytical essay composition across three consecutive academic semesters. Facilitated weekly discussion sections, evaluated research assignments, and assisted senior faculty with assessment rubrics.

Formal Education

BSc in Computer Science and Engineering

CGPA 3.69 / 4.00 · 117 credits completed
North South University, Dhaka • Mar 2022 – Dec 2024

Consistently on the Dean's List for Academic Excellence. Core coursework completed: Artificial Intelligence, Data Structures & Algorithms, Database Systems, Computer Vision, Distributed Systems, Natural Language Processing, Linear Algebra, and Discrete Mathematics.

Higher Secondary Certificate (HSC) — Science

GPA 5.00 / 5.00 (Perfect Score)
Govt. Mohammadpur Model School & College, Dhaka • 2019 – 2021

Recognition & Honors

ECE Capstone Project Showcase — 2nd Runner Up (12,000 BDT) · Innovation Challenge 262, Season 20. Selected from across the Department of Electrical and Computer Engineering for innovation, execution, and societal impact.

ECE Capstone Showcase Season 20 — 2nd Runner Up with Vice Chancellor and faculty

With Vice Chancellor · NSU Auditorium

2nd Runner Up — Project Showcase (Season 20)

12,000 BDT
Team receiving 2nd Runner Up check at ECE Capstone Season 20

Project Showcase Stage — Team Award Presentation

Department of ECE · Aligned with Innovation Challenge 262 · NSU Grand Auditorium.

National Hackathon

Solvio AI Hackathon group

Solvio AI

Top 10 Finalist (2025)

Autonomous agentic workflow for real-time multi-agent reasoning and orchestration.

Also Recognized

Top 20 — InnovativeX BUBT

Disaster-resilience platform AegisFlow with PostGIS spatial triage.

ICPC Asia Dhaka Regional Contestant (’22, ’23)

Graduate Research Aspirations

I am actively preparing for graduate studies (PhD / MSc by Research) in Computer Science. My primary research interests lie at the intersection of:

1. Safety-Critical NLP

Developing verifiable refusal boundaries and post-generation audit sinks for high-stakes advisory in developing economies.

2. Grounded Multimodal Vision

Mitigating VLM textual confirmation bias and diagnosing model capabilities on non-Latin, historical, and base-16 numerical scripts.

3. Efficient Edge Adaptation

Quantized sub-1B parameter models fine-tuned with multi-task structured supervision for offline edge deployment.

Academic Reference Letters

Confidential academic letters of recommendation from faculty supervisors at North South University (including Department of Computer Science and Department of History & Philosophy faculty) are readily available for official graduate admissions applications.

Let's discuss research & collaboration

Open to graduate research discussions, preprint feedback, and benchmark inquiries. Replies within 48 hours.