N. Tafannum

Front-end engineer · Dhaka · Remote

Nishat Tafannum

I build fast, accessible web applications with React, Next.js and TypeScript, and have done for 3+ years, for international ad tech and fintech clients. I lead design systems and shared component libraries, and work directly with client teams in fully remote settings. Alongside engineering, I research how language models handle Bangla.

See my work Résumé (PDF)

Selected work

Improve Digital for Azerion

Software Engineer L-2, Vivasoft · Apr 2026 – present

Front-end modernisation of an ad tech console, and a shared UI layer for its sister platforms.

  • Moved the build from Webpack to Vite and npm to Bun, cutting build time by 75%
  • Wrote 49 shared React building blocks, including a data table that replaced jQuery DataTables in 15+ modules, while cutting the codebase by a net ~7k lines
  • Started and lead a component library now shared by 3 platforms, with a design system whose tokens are enforced by 13 automated accessibility and style tests

React · TypeScript · Vite · Bun · Tailwind CSS · Ant Design · GitLab CI

NetCapital in-house at Azerion

Software Engineer L-1, Vivasoft · Aug 2025 – Mar 2026

Front-end architecture for Azerion’s in-house invoice financing platform, a multi-tenant system where merchants, buyers and financing providers manage high-value transactions.

  • Built the financing-order lifecycle, application approval, multi-account management and settlement workflows
  • Turned double-entry accounting, fee structures and product configuration into type-safe interfaces for factoring, loans and trade finance

Next.js 15 · TypeScript · RTK Query · NextAuth.js · Keycloak · Highcharts

PiHR HR & payroll SaaS

Software Engineer L-1, Vivasoft · May – Oct 2024

Reporting and forms across the HR modules of a micro-frontend product.

  • Integrated 60+ reports through a micro-frontend architecture with shared components
  • Added English, Bangla and Arabic support, including right-to-left layouts, without breaking existing views

Next.js · TypeScript · RTK Query · Zod · Tailwind CSS

VivaKits open-source component library

Software Engineer L-1, Vivasoft · Mar – Jun 2024

Accessible React components (Input, Select, OTP, Skeleton, Switch) with full API documentation, published on npm.

React · TypeScript · Tailwind CSS

Path

The path so far
The degree branches into engineering, where I’ve worked since 2023, and research, which grew out of my thesis. Footprints mark the path walked so far and dashes the road ahead; projects are places along each track, and brackets show the role each engineering project was done in. Time runs left to right (top to bottom on a phone), with 2018–2022 compressed. Select anything to read about it below.

Skills

Front end
  • React
  • Next.js
  • TypeScript
  • JavaScript
  • Tailwind CSS
  • Ant Design
  • shadcn/ui
State & data
  • Redux Toolkit
  • RTK Query
  • React Query
  • Zod
  • Highcharts
  • ApexCharts
Architecture
  • Design systems
  • Component libraries
  • Micro-frontends
  • Role-based access control
Build & delivery
  • Vite
  • Bun
  • tsup
  • GitLab CI
  • Docker
  • Private npm registry
Quality
  • WCAG AA accessibility
  • ESLint
  • Biome
  • Automated style & contrast tests
Auth
  • NextAuth.js
  • Keycloak
ML & NLP
  • Python
  • PyTorch
  • Hugging Face Transformers
  • scikit-learn
  • CLIP
  • BanglaBERT
Workbench
  • Git
  • Claude Code
  • Jupyter

Research

Alongside engineering, I research bias and authorship in Bangla and other low-resource languages. The research tracks on the map have the details.

  1. Aupi, M. R. I., Tafannum, N., Rahman, M. S., Hassan, K. M., & Rahman, N. (2025). WoNBias: A dataset for classifying bias and prejudice against women in Bengali text. Proceedings of the 6th Workshop on Gender Bias in Natural Language Processing (GeBNLP), co-located with ACL 2025. paper
  2. Anisujjaman, M., Tafannum, N., Ghosh, B., Haque, M. S., & Islam, T. (2026). Multimodal semantic image clustering through adaptive fusion of visual and textual representations. Submitted to a conference.
  3. Tafannum, N. (2026). Human vs AI-generated text classification using multi-criteria model selection and LIME-based explainability. Manuscript. code

Certificates

DataCamp · Python & ML · 2026
DataCamp · Infrastructure · 2026

Languages

English · IELTS 7.5

Bangla · Native

Contact

The best way to reach me is by email at sejuty87@gmail.com. I’m open to front-end and full-stack roles, remote or in Dhaka.

Résumé (PDF) · GitHub · LinkedIn