Tanvir Siddique
Statement of Purpose • Direction
~3 min read 650 words

Building, Learning, and What Comes Next

First-year CSE student at KUET. Building practical tools, open-source automation, and learning ML foundations.

Jump to Section ▾
"If there's a repetitive manual task, a messy dataset, or a real problem students face, I sit down, learn what's needed, and write the code to solve it."

1. Who I Am: Building & Contributing to Open Source

I am currently a first-year undergraduate in Computer Science & Engineering at Khulna University of Engineering & Technology (KUET). My programming journey has always been about learning by building real, useful things.

During my college years at Notre Dame College, Dhaka, as part of the Notre Dame English Club Wikipedia Editorial & Research Team (NDEC WERT), I started writing Python scripts using Pywikibot to automate community operations. Since then, I've expanded to large-scale infrastructure and localization:

  • Nuvio Streaming Ecosystem: Engineered native Bengali (bn) localization across the official NuvioMobile and NuvioTV Kotlin repositories, bringing accessibility to over a million active users on Android and Android TV platforms.
  • Wikimedia Tool Developer: Built CampAnalytics for Project Korikath—a production analytics tool deployed on Wikimedia Toolforge and Streamlit that tracks campaign participation and contributor retention.
  • Global Bot Operations: Secured cross-wiki automated editing permissions, completing over 47,000 automated bot edits and 35,000+ media file operations across Wikipedia, Wikidata, and Wikimedia Commons. Structured roughly 10,000 articles on Fulfulde Wikipedia to map core categories from English Wikipedia.

2. Developer Tooling: Building CodeFunc

I like tools that are lightweight and stay out of the way. When inspecting unfamiliar codebases or checking PRs, jumping between files and documentation creates friction.

To solve that, I built and published CodeFunc (available on the VS Code Marketplace and Chrome Web Store):

  • Line 0 File Summary: Uses VS Code's CodeLens API to show a concise summary and structure at the top of each file without opening side panels.
  • AI Code Explanations: Connects to Gemini, Groq, OpenRouter, and Claude for contextual code breakdowns.
  • Local AST Parsing: Works offline using regex and AST heuristics for instant signature and export analysis.

3. Shipping Under Pressure: CampusShare KUET

During the 50-hour ForkedArch Freshers Hackathon 2026 (Forkathon 2026), our team secured 1st Place building CampusShare KUET. We tackled a common problem in KUET residential halls: "Does anyone have an Arduino kit or a scientific calculator right now?":

  • Full-Stack Setup: Built an asynchronous FastAPI backend with PostgreSQL on Render, paired with a React 18 web app.
  • Handover Verification: A dynamic 6-digit OTP and QR code system ensuring items are verified in person before marking as borrowed or returned.
  • Cross-Platform Build: Packaged the app with Capacitor into an installable Android APK for students.

4. Applied Machine Learning

I am actively studying machine learning and data science, aiming to understand the underlying mathematics rather than just calling pre-packaged APIs:

  • AgroScan & Smart Farm: Trained a 38-class plant pathology classifier on leaf photos as AgroScan. Contributed the model into production for the smart-farm platform (React + Supabase).
  • Language Modeling (MLP-Implementation): Built a character-level Bigram Language Model in PyTorch on The Wizard of Oz text—mapping character vocabularies, calculating cross-entropy loss, and sampling text with temperature.
  • Competitions: Participating in the University of Tokyo Matsuo-Iwasawa Lab GCI World 2026 program and Kaggle challenges to practice feature engineering and gradient boosting.

5. Where I Am Heading: Systems & Architecture

As I progress through my CSE coursework at KUET, my goals are focused on deeper technical layers:

  • Algorithmic Problem Solving: Practicing core data structures, graph algorithms, and competitive programming problems in C++ to sharpen analytical problem-solving speed.
  • Systems Fundamentals: Learning operating systems, memory management, and data structures to write highly efficient software.
  • Autonomous Multi-Agent Systems: Integrating and orchestrating local LLM pipelines and automated agents (via Google Antigravity and Ollama) for workflow automation and repository maintenance.

6. Working Together

If you are looking for an intern, mentee, or collaborator:

  • Fast Learner: When I encounter a new problem or stack, I dive in, read the documentation, and build a working solution.
  • Hands-On Experience: Comfortable working across Python, TypeScript, React, Kotlin, C++, and basic ML workflows.
  • Work Ethic: Early in my degree, but highly motivated, open to feedback, and dedicated to writing clean, maintainable code.

I am eager to take on software engineering internships, research opportunities, and open-source collaborations. Feel free to reach out.