DEV Community: Samaina Khatun

SlothDesk: Make Room for Slow with Open-Source AI

Samaina Khatun — Fri, 23 Oct 2026 14:59:30 +0000

SlothDesk: Make Room for Slow

Productivity tools often make people feel like they are falling behind.

SlothDesk takes a different approach: it turns an overwhelming task list into a calm, achievable plan made of tiny wins.

What I built

SlothDesk is a gentle task planner designed around the Week 3 “Slow and Steady” theme.

It includes:

A simple daily task planner
Tiny-win task labels
Task completion tracking
A progress indicator
A calm, low-pressure interface
Local-first AI positioning
Support for connecting to open-source models through Ollama or LM Studio

Why open-source AI?

The goal is to keep the planning experience private and accessible.

SlothDesk is designed to work with an OpenAI-compatible local model, such as an Ollama-hosted model. This means users can eventually process their task lists locally instead of sending personal productivity data to a hosted service.

The .env.example file includes the configuration needed to connect a local model.

How it fits “Slow and Steady”

SlothDesk is intentionally not a hustle dashboard.

It does not focus on:

Streaks
Productivity scores
Competitive rankings
Guilt-driven reminders

Instead, it encourages people to start with one small task and build momentum gradually.

The interface is based on three principles:

Break large tasks into tiny steps.
Keep the experience human and pressure-free.
Let users own their data.

Tech stack

Next.js
React
TypeScript
CSS
OpenAI-compatible local model support
Ollama and LM Studio compatibility

Run locally


bash
git clone https://github.com/samainakhatun115-cpu/slothdesk.git
cd slothdesk
npm install
npm run dev

HustleCraft AI: Local-First Open-Source Invoice Intelligence for Freelancers

Samaina Khatun — Wed, 14 Oct 2026 14:54:28 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

HustleCraft AI is an open-source, local-first invoice and expense intelligence tool built specifically for freelancers, independent contractors, and solo side-hustlers.

Freelancers waste hours each month sorting disorganized email receipts, re-typing invoice details, and guessing quarterly taxes. Existing accounting software charges steep monthly subscription fees ($30–$60/month) and forces sensitive client billing records onto proprietary cloud databases.

HustleCraft AI runs 100% locally on-device. It solves three critical problems:

Automated Invoice Structuring: Ingests messy invoice drafts or client emails and extracts client names, line items, and totals into clean, tax-ready records.
Expense & Tax Deduction Classification: Automatically sorts receipts into tax-writeoff categories (SaaS, Hardware, Office, Client Meetings).
Quarterly Tax Set-Aside Projections: Computes net revenue after expenses and estimates recommended quarterly tax reserves so solopreneurs are never surprised during tax season.

Who Would Pay For It?

Freelancers and solo digital agency owners. Instead of paying continuous monthly SaaS fees to cloud accounting platforms, freelancers gladly pay a one-time license fee ($29–$49) for a desktop utility that preserves complete client confidentiality and eliminates recurring subscription costs.

Demo

Here is the local CLI execution flow:

[HustleCraft AI] Initialized Local Engine [Model: Gemma-2-9B / Llama-3.1-8B Local]

--- 1. Parsing Client Invoice ---
{
  "invoice_id": "NS-8921",
  "client": "NorthStar Design Agency",
  "total_usd": 2850.0,
  "status": "Ready to send",
  "extracted_at": "2026-10-14 20:23:00",
  "privacy_mode": "100% On-Device / Zero Cloud Storage"
}

--- 2. Auditing Business Receipt ---
{
  "category": "Software & SaaS",
  "amount_usd": 48.0,
  "tax_deductible": true,
  "notes": "Auto-classified under Software & SaaS for year-end Schedule C deduction."
}

--- 3. Cashflow & Quarterly Tax Projection ---
{
  "total_revenue_usd": 2850.0,
  "total_expenses_usd": 48.0,
  "net_profit_usd": 2802.0,
  "quarterly_tax_setaside_usd": 700.5,
  "take_home_usd": 2101.5
}

Code

The project is fully open source under the MIT License:

📂 GitHub Repository: https://github.com/samainakhatun115-cpu/hustlecraft-ai
🚀 Tech Stack: Python 3.9+, Open-Weight LLMs (Gemma 2 / Llama 3.1 via Ollama), Regex & JSON Structuring.

How I Built It

HustleCraft AI is built around open-weight model architectures (such as Google Gemma 2 and Meta Llama 3.1) executing locally via Ollama and Hugging Face:

flowchart TD
    User["Freelancer / Solopreneur"] -->|Drops Raw Invoices / Receipts| App["HustleCraft AI"]
    App --> Parser["Local Invoice Parser"]
    App --> Categorizer["Receipt & Expense Auditor"]

    Parser -->|Prompt Chunks| LocalLLM["Open-Weight LLM (Gemma 2 / Llama 3.1 / Ollama)"]
    Categorizer -->|Prompt Chunks| LocalLLM

    LocalLLM --> Engine["Cashflow & Tax Planner"]
    Engine --> Output["Clean JSON & Tax-Ready Summary"]
    Output --> User

Extraction Pipeline: Uses constrained prompt templates to parse messy client correspondences without hallucinations.
Deterministic Fallbacks: Employs rigorous regex and rule validations to guarantee numerical accuracy for financial totals.
Zero-Cloud Dependency: Designed so no internet connection is required after downloading the model weights.

Why Does Open Innovation Matter?

Open innovation is fundamental to why this product works:

Total Client Privacy: Freelancers deal with non-disclosure agreements (NDAs) and proprietary client rates. Sending client invoices to closed proprietary APIs risks leaking commercial terms. With open-weight models, client data never leaves the laptop.
Economic Viability for Solo Businesses: A side hustle cannot afford per-token API charges or monthly minimums that eat into slim profit margins. Open weights mean the marginal cost per invoice parsed is exactly $0.
No Risk of Vendor Deprecation or Lock-in: Closed AI APIs frequently change pricing, deprecate model versions, or alter system prompts. Open-weight models guarantee stability and perpetual ownership.

My First User

I tested HustleCraft AI with a freelance UI/UX designer friend who was drowning in three months of unfiled receipts before quarterly tax deadlines.

She ran HustleCraft AI over 18 messy client PDFs and invoices:

Her reaction: "The fact that I don't have to upload my client contracts and billing rates to a random cloud server makes this an immediate keeper. I'd pay $30 upfront for this instead of paying QuickBooks every single month."
Requested next feature: Auto-generating a shareable PDF invoice directly from the parsed data.

Prize Categories

Overall Week 2 Winner

AgentCraft AI: An Open-Source Multi-Agent Workspace Powered by Open-Weight LLMs

Samaina Khatun — Sat, 10 Oct 2026 17:36:49 +0000

This article is a submission for the Hacktoberfest Open-Source AI Challenge: Week 1.

🌟 Inspiration & Background

Building with AI shouldn't require sending private codebase architecture or sensitive developer documentation to proprietary cloud APIs.

With the rapid advancement of open-weight AI models (such as Google's Gemma 2, Meta's Llama 3, and local runtime tools like Ollama and Hugging Face Transformers), developers now have the power to run high-performance AI agents completely offline on local hardware.

I created AgentCraft AI — a lightweight, modular multi-agent framework designed to orchestrate specialized AI agents for code auditing, documentation synthesis, and project planning, all powered by open-weight AI.

✨ What It Does

AgentCraft AI provides three specialized developer agents:

🏗️ Code Architect Agent: Inspects code structure, identifies methods/classes, audits security risks, and suggests architectural refactoring.
📑 Docs Summarizer Agent: Chunks and synthesizes lengthy technical specifications or API docs into executive summaries and bullet points.
🎯 Task Planner Agent: Decomposes high-level developer goals into step-by-step implementation milestones and generates dynamic Mermaid workflow diagrams.

🏗️ Architecture & Multi-Agent Flow

flowchart TD
    User["Developer"] -->|Submits Code / Task Goal| AgentCraft["AgentCraft AI Workspace"]
    AgentCraft --> Architect["Code Architect Agent"]
    AgentCraft --> Summarizer["Docs Summarizer Agent"]
    AgentCraft --> Planner["Task Planner Agent"]

    Architect -->|Local Inference| OpenWeightLLM["Open-Weight LLM (Gemma 2 / Llama 3 / Ollama)"]
    Summarizer -->|Local Inference| OpenWeightLLM
    Planner -->|Local Inference| OpenWeightLLM

    OpenWeightLLM -->|Parsed Output| Results["Structured JSON & Mermaid Diagrams"]
    Results --> User

💻 Open-Source Code & Setup

The project is 100% open-source under the MIT License.

📂 GitHub Repository: samainakhatun115-cpu/agentcraft-ai
🚀 Tech Stack: Python 3.9+, Open-Weight Models (Gemma 2 / Ollama / Hugging Face), Standard JSON Parsers.

Running AgentCraft AI:

from app import AgentCraft

# Initialize Workspace with an open-weight model engine
framework = AgentCraft(model_name="Gemma-2B / Llama-3-OpenWeight")

# 1. Run Code Architect Agent
code_analysis = framework.architect_agent(code_snippet)

# 2. Run Docs Summarizer Agent
summary = framework.summarizer_agent(doc_text)

# 3. Run Task Planner Agent
plan = framework.planner_agent("Deploy Local AI Code Auditor")

🎯 Hacktoberfest 2026: AI Belongs to Everyone

Hacktoberfest 2026's central mission — "AI belongs to everyone" — highlights the crucial role of open-source models and toolings. By running multi-agent workflows with open-weight models, developers maintain complete privacy, transparency, and freedom from vendor lock-in.

🚀 What's Next

🌐 Web UI Dashboard built with Streamlit / Gradio.
🛠️ Agentic Function Calling & Local File System Execution.
💬 Local Chat Interface for interactive Q&A.

Thank you for reading, and happy Hacktoberfest! 🎃✨

BuddyCraft AI: An Open-Source AI Study Companion Built for a Friend

Samaina Khatun — Fri, 02 Oct 2026 14:38:58 +0000

This article is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend.

🌟 Inspiration

Studying long, dense technical documentation, scientific papers, or complex API references can quickly lead to cognitive overload and focus fatigue.

Recently, my friend was preparing for intensive exams while juggling open-source projects, and constantly felt overwhelmed by dense academic literature. Traditional reading felt slow, and generic AI tools lacked a supportive, study-focused structure.

I built BuddyCraft AI — an open-source, AI-powered study companion designed specifically to break down dense reading materials into gentle summaries and self-testing interactive flashcards.

✨ What It Does

BuddyCraft AI takes any input text or technical document and:

📑 Generates Key Concept Summaries: Distills long paragraphs into concise, digestible takeaways.
🎴 Auto-Generates Q&A Flashcards: Extracts core statements and turns them into self-testing study cards.
🤖 Leverages Open-Weight AI Models: Built to run seamlessly on top of open-weight models (Hugging Face / Ollama / Llama 3 / Gemma) to maintain privacy and customization.
💙 Provides a Supportive Persona: Acts as an encouraging study partner to keep motivation high during late-night study sessions.

🏗️ Architecture & Data Flow

flowchart TD
    User["Friend / Student"] -->|Inputs Reading Material| UI["BuddyCraft Interface"]
    UI -->|Sends Text Chunks| PromptEngine["Prompt & Extraction Engine"]
    PromptEngine -->|Inference Call| OpenWeightLLM["Open-Weight Model (Ollama / HF / Gemma)"]
    OpenWeightLLM -->|Returns Key Statements & Q&A| OutputParser["JSON & Structure Parser"]
    OutputParser -->|Flashcards & Summaries| UI
    UI -->|Interactive Study Session| User

💻 Code & Repository

The project is fully open-source under the MIT License.

📂 GitHub Repository: samainakhatun115-cpu/buddycraft-ai
🚀 Tech Stack: Python, Open-Weight AI Pipelines, Standard JSON Output Parsers.

Sample Usage:

from app import summarize_text, generate_flashcards

text = "Open-weight AI models allow developers to inspect, fine-tune, and deploy machine learning models on private infrastructure without vendor lock-in."

# 1. Summarization
summary = summarize_text(text)
print("Summary:", summary)

# 2. Flashcard Generation
cards = generate_flashcards(text)
for card in cards:
    print(f"Q: {card['question']}\nA: {card['answer']}")

🎯 Hacktoberfest 2026 & Open-Source AI

Hacktoberfest 2026's theme "AI belongs to everyone" resonates deeply with this project. Building tools with open-weight models ensures that educational AI assistants remain accessible, inspectable, and customizable by anyone without being locked into proprietary APIs.

🚀 What's Next

🎙️ Adding Text-to-Speech audio support for auditory learners.
📦 Creating a lightweight Web UI (Streamlit / Gradio) for instant browser use.
🤝 Expanding open-weight model fine-tuning for specific academic disciplines.

Thank you for reading, and happy Hacktoberfest! 🎃✨

introduction

Samaina Khatun — Thu, 01 Oct 2026 17:27:23 +0000

hi I am new to the community,happy to be a part of dev

DEV Community: Samaina Khatun

SlothDesk: Make Room for Slow with Open-Source AI

SlothDesk: Make Room for Slow

What I built

Why open-source AI?

How it fits “Slow and Steady”

Tech stack

Links

Run locally

HustleCraft AI: Local-First Open-Source Invoice Intelligence for Freelancers

What I Built

Who Would Pay For It?

Demo

Code

How I Built It

Why Does Open Innovation Matter?

My First User

Prize Categories

AgentCraft AI: An Open-Source Multi-Agent Workspace Powered by Open-Weight LLMs

🌟 Inspiration & Background

✨ What It Does

🏗️ Architecture & Multi-Agent Flow

💻 Open-Source Code & Setup

Running AgentCraft AI:

🎯 Hacktoberfest 2026: AI Belongs to Everyone

🚀 What's Next

BuddyCraft AI: An Open-Source AI Study Companion Built for a Friend

🌟 Inspiration

✨ What It Does

🏗️ Architecture & Data Flow

💻 Code & Repository

Sample Usage:

🎯 Hacktoberfest 2026 & Open-Source AI

🚀 What's Next

introduction