Mohammad Borhan Uddin
github·linkedin·brhn.contact@gmail.com·blog
November 2026
1Introduction
I am a computer science graduate who builds full-stack products with React, Next.js, Node.js and TypeScript, with a focus on AI integration. I care most about software that people quietly come to rely on.
2Toolchain
Table 1 lists the tools used across the projects in §3 and the thesis in §5.
| Area | Tools |
|---|---|
| Languages | TypeScript, JavaScript, Python, C++, PHP, SQL |
| Frontend | React, Next.js, Tailwind CSS, shadcn/ui, Material UI, Bootstrap |
| Backend | Node.js, Express.js, REST APIs, Auth.js, Passport.js, Zod |
| Data | MySQL, MongoDB, Firebase, Prisma, Mongoose |
| AI / ML | Scikit-learn, YOLO, LangChain, CrewAI |
| Tools | Git, GitHub, Docker, Stripe, Vercel, Postman |
Table 1: Tools used in this work, by area.
3Selected Work
Five systems are described below. The first four have a case study with a video and screenshots.
3.1ObhijogNama
▶A university complaint portal with role-based access, assignment, tracking and real-time notifications. Impact: Selected by NSTU for university-wide use.
React · Express.js · MySQL · Firebase · Gemini API · shadcn/ui
3.2NSTU Cover Page Generator
▶An installable PWA that formats lab-report cover pages and indexes in seconds. Impact: 7,000+ students every month since 2022.
JavaScript · Tailwind CSS · PWA
3.3ShopVerse
▶A full e-commerce platform with Stripe checkout and media stored on IPFS.
Next.js · TypeScript · Prisma · MySQL · Stripe · Pinata (IPFS)
3.4PetSoft
▶A subscription-based pet management SaaS.
Next.js · TypeScript · Auth.js · Prisma · Zod · Stripe · shadcn/ui
3.5Advanced Task Manager
4Education and Problem Solving
- CGPA: 3.60 / 4.00
- Relevant coursework: Data Structures, Algorithms, Software Engineering, Database Systems, Artificial Intelligence, Machine Learning
Alongside the degree I competed in algorithmic programming; Table 2 lists the problems I have solved and my peak ratings.
| Judge | Handle | Solved | Peak rating |
|---|---|---|---|
| Codeforces | borhan08 ↗, borhan_08 ↗ | 799 | 1360 (Pupil) |
| LeetCode | borhan008 ↗ | 31 | 1536 (contest) |
| CodeChef | borhan008 ↗ | 20 | 1880 (4★) |
| VJudge | ASH2101008M ↗ | 69 | — |
| LightOJ | borhan008 ↗ | 14 | — |
| Timus | borhan008 ↗ | 11 | — |
| Gonitzoggo | borhan ↗ | 18 | — |
| Total | 962 | ||
Table 2: Problems solved by judge, November 2026.
- Champion, Junior Segment, NSTU Intra-University Programming Contest (CSTE Club), 2022.
- Champion, Junior Segment, NSTU Intra-University Programming Contest (ICE Club), 2022.
- 3rd Place, NSTU Intra-University Individual Programming Contest, 2022.
- 6th Place, NSTU Intra-University Programming Contest, 2023.
5Undergraduate Thesis: Bangla Fact Verification
A team project, supervised by Iftekhar Mahmud Towhid, Assistant Professor, Dept. of CSTE, NSTU. 2026.
Our undergraduate thesis [1] asks whether a model can verify a Bangla news item by reading its text and looking at its image. We proposed LACR-Fusion, a multimodal model, and compared it against a text-only BanglaBERT, an image-only EfficientNet-B0, and simple fusion baselines (Figure 1).
Figure 1: The LACR-Fusion pipeline. Text and image are encoded separately, fused, and mapped to a calibrated verdict.
5.1Results
On 9,600 Bangla news text–image pairs, the fused model reaches
and post-hoc calibration reduces the expected calibration error from
However, leakage audits and missing- and mismatched-image stress tests show that every model relies mainly on the text and gains little from the image, a finding that matters for anyone building multimodal fact-checkers.
5.2Evidence corpus
To support the study we built a distributed evidence-collection and annotation platform, modelled on FEVER and AVeriTeC, with role-based access, independent multi-reviewer verification and adjudication. It collected 21,133 articles from 72 Bangla news sources.
5.3Research Interests
Looking ahead, my interests extend beyond fact verification to bioinformatics, computational biology and psychology. What draws me to these fields is the question that ran through this thesis: how to build models that learn reliably from noisy, human-centred data, and that stay honest about what they do not know.
I am actively looking for opportunities (research assistantships, graduate study and collaborative projects) where I can contribute, learn and grow. If you work in these areas, I would be glad to hear from you at brhn.contact@gmail.com.
Correspondence
Write to brhn.contact@gmail.com.
References
- [1]M. B. Uddin, M. Raihan, and F. I. Mily, “Evidence-Grounded Multimodal Fact Verification for Bangla News,” B.Sc. (Engg.) thesis, supervised by I. M. Towhid, Noakhali Science and Technology University, 2026.
- [2]M. B. Uddin, Bangla Dictionary. Chrome extension: double-click any English word to see its Bangla meaning. github.com/borhan008/ChromeExtensions/tree/main/bangla-dictionary
- [3]M. B. Uddin, Heart Disease Prediction API. ML model served from a Docker container. github.com/borhan008/heart-model-docker
- [4]M. B. Uddin, People Flow Detection. Detecting and tracking people with YOLO. github.com/borhan008/yolo-People-Flow-Detection-and-Tracking
- [5]M. B. Uddin, CSS Grid Bangla. A CSS Grid tutorial written in Bangla. borhan008.github.io/css-grid-bangla
- [6]M. B. Uddin, Halal Audios Bangla. An audio library web app. halalaudios.netlify.app/index.html
- [7]M. B. Uddin, Anonymous Compliments. Create a link, receive anonymous compliments. anonymousmessage.web.app
- [8]M. B. Uddin, Department Website. Frontend for a university department site. nstu-beta.netlify.app