Projects

A selection of technical and business projects - shipped tools, automation scripts, and competition work.

Bangladesh Youth Tax Calculator logo
AI Skill Β· Tax Tech

Bangladesh Youth Tax Calculator

An AI Agent Skill and deterministic statutory engine made for students, interns, and young professionals filing on etaxnbr.gov.bd.

Having an e-TIN, filing tax returns is something we all face. But being a student with no budget to hire a CA firm, many turn to AI. Generic models struggle with the NBR portal UI and hallucinate balance sheet math.

This skill equips your LLM with the Income Tax Act 2023, guides you through every screen, and uses a deterministic Python engine to ensure your balance sheet difference is exactly 0.00 while keeping tax expense at minimum.

View Project

Tools & Standards

AI Agent SkillPython 3 EngineIncome Tax Act 2023etaxnbr.gov.bdForm IT-10BMIT License

Core Outcomes

βš–οΈKeeps tax expense at minimum (0 BDT floor under 4,00,000 BDT)
πŸ’΅100% bank TDS refund with direct claiming guidance
πŸ“‘Zero-difference balance sheet math via Section 56(g)
πŸ›‘οΈPrevents NBR e-Return portal errors & audit flags

Standalone Engine

$python calculator.py

Open source on GitHub with 1-click installation for Windows and macOS.

github.com/zaifears/youth-tax-calculator
Tapo-Viewer logo
Windows Β· Camera Viewer

Tapo-Viewer

A Windows desktop app for playing live Tapo camera feeds and downloading recorded clips directly from a camera's MicroSD card.

What it does

πŸ“ΊLaunches local RTSP live streams in VLC, mpv.net, PotPlayer, or FFplay
πŸ“…Browses MicroSD recordings with a calendar and motion-event timeline
πŸ“₯Downloads clips with MP4 conversion and a persistent download manager
πŸ”’Communicates directly with the camera over home Wi-Fi/LAN
Tapo desktop app for Windows

Windows 10/11 Β· Python 3.10+ Β· MIT License Β· Powered by PyTapo

Download standalone .exe
LocReminder screenshot 0
LocReminder screenshot 1
LocReminder screenshot 2
LocReminder screenshot 4
LocReminder app icon
Flutter Β· Android

LocReminder

A location-based alarm for Android, which helps you wake up at the right place.

I kept missing my stop on long bus rides in Dhaka, and a normal alarm cannot help with that since you never know exactly when you will arrive.

LocReminder rings a real alarm when you reach a place you pinned, not at a set time. Sound, vibration and a full screen alert, even on silent and a locked screen.

No accounts, no analytics, no server. Every release is signed and scanned by VirusTotal before it ships.

View Project

Tools used

FlutterDartKotlinOpenStreetMapGitHub Actions

What it does

πŸ””Wakes you with a real alarm, not a quiet notification
πŸ—ΊοΈRuns on OpenStreetMap, no API key or Google account
πŸ”‹Battery-aware polling so a long trip won't drain your phone
πŸ”’Your location never leaves your device

Published on GitHub Releases, with an F-Droid submission in progress.

Our department was tracking absences for 9 teams in a single shared spreadsheet, no access control, no validation, pure chaos.

My first idea was to fix it entirely with Excel formulas. That lasted about 10 minutes.

So I learned VBA from scratch and built a proper role-based HR absence management system, in a single day.

Watch on YouTube

Tools used

Excel VBAUserFormsClass ModulesXLOOKUPSUMIFSINDIRECT

What it does

πŸ”Role-based login, 9 teams, 9 passwords, each team sees only their own data
⚑Auto-fill, select a name, ID and designation populate instantly
πŸ“…Custom calendar date picker, built from scratch in VBA, no third-party tools
πŸ“ŠIndividual report generator, one click pulls any employee's full absence history
πŸ”„Live leave totals, automatically aggregated across all 9 teams
πŸ’ΎAuto-save, data saves silently every time a user leaves their sheet
πŸ–₯️Looks and feels like a standalone app, no Excel ribbon, no sheet tabs, no formula bar

Built entirely in Microsoft Excel VBA - no database, no server, no paid tools.

Python Β· Automation

AML/CFT Scraper Utility

Investigating unauthorized "Diamond Top-up" platforms was a highly repetitive task. I had to click through dozens of sketchy e-commerce sites manually just to find which personal bKash accounts were being illicitly used as payment gateways.

To speed up my own investigation workflow, I built a stealth-enabled automation script that navigates these dynamic sites, extracts the hidden numbers, and logs the necessary visual evidence.

It turned a tedious multi-minute process per site into a 10-second automated batch sweep. I packaged it into a standalone executable so it runs perfectly on my locked-down office laptop without needing admin rights.

View Source Code

Tools used

Python 3PlaywrightTkinterconcurrent.futuresOpenPyXLPyInstaller

What it does

πŸ•΅οΈβ€β™‚οΈStealth Navigation, bypasses basic anti-bot protections to handle complex SPAs and accordions
πŸ“ΈForensic Logging, automatically captures and names screenshots at crucial transaction steps
⚑Batch Processing, scans multiple flagged URLs simultaneously using multi-threading
πŸ“ŠAuto-Reporting, compiles findings directly into a clean, auto-generated Excel database
πŸ“¦Standalone Executable, compiled to a zero-dependency .exe to run on restricted corporate laptops
πŸ›‘Manual Bridge, allows pausing the script to manually bypass CAPTCHAs before resuming

Built strictly to optimize personal investigation workflows - standalone Python utility.