DEV Community: Kothakula Navadeep

SideGig AI: Your Privacy-First Financial Assistant for Freelancers

Kothakula Navadeep — Sun, 18 Oct 2026 15:18:21 +0000

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

What I Built

I built SideGig AI, a privacy-first financial assistant for freelancers and small business owners.

Managing receipts, invoices, and expenses manually takes time. SideGig AI aims to simplify this process by helping users extract invoice details, categorize expenses, detect potential duplicate invoices, visualize spending, and export financial records as CSV files.

The goal is to make financial management easier for freelancers without requiring expensive AI subscriptions.

Potential users include freelance developers, designers, content creators, and small business owners who want to spend less time on paperwork.

Demo

Screen Recording 2026-10-18 195356.mp4 - Google Drive

drive.google.com

Code

NavadeepKothakula / sidegig-aipro

<<<<<<< HEAD

SideGig AI 💼🤖

A Privacy-First, Open-Weight AI Finance Assistant for Freelancers, Creators & Side Hustlers. Built for the Hacktoberfest Open-Source AI Challenge — Week Theme: Side Hustle.

🚀 Overview & Vision

Freelancers, independent developers, creative contractors, and students juggling side hustles receive bills, invoices, receipts, and payment receipts scattered across emails, PDFs, screenshots, and paper scans. Manually transcribing numbers into spreadsheets is slow, prone to errors, and tedious. At the same time, commercial AI accounting software often requires expensive monthly subscriptions and uploads private personal financial documents to centralized cloud servers.

SideGig AI is designed from the ground up around a zero-cloud-fees, local-first, privacy-by-default architecture:

100% In-Browser Document Extraction: Parse native text from PDFs using PDF.js and extract text from receipts/images using in-browser Tesseract.js OCR. No document files are ever sent to external cloud APIs.
Deterministic Financial Accuracy: Uses integer minor units (cents) to mathematically…

View on GitHub

How I Built It

I built SideGig AI using React, TypeScript, Vite, and Tailwind CSS.

The project uses open-source document-processing tools such as Tesseract.js for OCR and PDF.js for PDF text extraction. IndexedDB provides local data persistence, while Recharts powers the expense analytics dashboard.

The AI architecture is designed to support local inference through Ollama and a compatible open-weight language model. This allows AI-powered expense categorization and financial insights without depending on a paid AI API.

For essential operations such as calculating expenses, filtering transactions, and exporting reports, the application can use deterministic application logic.

Why Does Open Innovation Matter?

Open innovation makes it possible to build useful AI-powered tools without making proprietary AI services a mandatory dependency.

Privacy: Local-first storage and local model inference can help users keep sensitive financial documents on their own computers.

Affordability: Running an open-weight model locally can eliminate per-request hosted AI charges, making frequent use more affordable.

Flexibility: Developers can experiment with different models and adapt the application to specific business needs.

Less vendor lock-in: A modular AI architecture gives developers more control over model selection and future changes.

For freelancers and small businesses, these advantages matter. They need affordable tools that solve real problems without introducing unnecessary costs or exposing sensitive information.

SideGig AI explores how open-source software and open-weight AI can make financial organization more accessible.

🌿 WildLens: Explore Nature, Not Your Screen.

Kothakula Navadeep — Sun, 11 Oct 2026 16:39:08 +0000

This is a submission for the "Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass" (https://dev.to/challenges/hacktoberfest-week1-2026-10-05)

What I Built

WildLens is an AI-powered nature companion designed to help people step away from their screens and explore the world around them.

The idea is simple: technology should encourage us to experience nature, not keep us glued to our devices.

With WildLens, users can:

🌱 Discover Nature: Capture a photo of a plant or bird and use AI to explore possible identifications.
🥾 Complete Outdoor Missions: Take on real-world challenges, such as identifying different leaf shapes, observing birds, or spending 15 minutes outdoors.
📖 Maintain a Nature Journal: Save discoveries, observations, and photographs to build a personal record of outdoor adventures.
📵 Explore Offline: Work toward an offline-first experience where AI inference and saved observations remain on the user's device.

WildLens is designed for nature enthusiasts, students, hikers, curious beginners, and anyone who wants to reconnect with the natural world.

The goal is to make the screen the shortest part of the experience: discover something, learn about it, and go outside to find more.

Demo

🎥 Video Demo: https://drive.google.com/file/d/1B0fO1NHYgujwa29I3714irAgs-Le1x-b/view?usp=drive_link

The project is intended to use an open-source web stack and locally executed AI inference, making it easier for other developers to inspect, modify, and extend.

How I Built It

WildLens is planned as an offline-first web application built around open AI models rather than a proprietary cloud-only AI API.

Technology stack:

React + TypeScript + Vite: For a responsive, maintainable web application.
Tailwind CSS: For a clean, mobile-friendly interface.
Transformers.js: To run compatible machine-learning models directly in the browser.
Open-weight vision model: A model such as "Xenova/clip-vit-base-patch32" can help compare an image with candidate descriptions and suggest possible matches.
ONNX Runtime: For browser-compatible inference where supported by the selected model and runtime.
IndexedDB: To store the user's nature journal and observations locally.
Progressive Web App (PWA): To support installation, caching, and offline access to supported features.

The AI is an important part of the experience, not an extra chatbot added to a conventional website. It helps users interpret what they discover outdoors, while missions turn those discoveries into real-world activities.

The intended workflow is straightforward:

Open WildLens and choose a nature discovery or outdoor mission.
Capture or upload an image.
Let the local AI model compare the image against candidate descriptions.
Review the suggested matches and learn more.
Save the observation locally and return to exploring.

A key technical consideration is that CLIP is a general-purpose image-and-text matching model, not a specialist bird or plant classifier. Its suggestions need to be treated as possible matches rather than guaranteed identifications.

Why Does Open Innovation Matter?

For WildLens, open innovation is about making AI more accessible, adaptable, and privacy-conscious.

Privacy by design

A local-inference architecture can process images on the user's device instead of sending every photograph to a third-party AI server. Local storage can also keep personal observations under the user's control.

Reduced dependence on paid APIs

Once the necessary model assets are available, supported inference can run locally without a per-request cloud inference charge. Users still need a compatible device, and initial model downloads may require internet access.

Freedom to experiment

An open, replaceable model pipeline makes it possible to test different vision models, compare their performance, and adapt the application to new use cases without redesigning the entire product around one provider.

Access beyond reliable internet

Nature exploration often happens in parks, gardens, hiking trails, and rural areas where connectivity is unreliable. Local inference and properly cached application assets can make supported features available offline.

Community-driven improvement

Developers can inspect the implementation, improve the identification pipeline, contribute new outdoor missions, and build on the project. Model licenses still matter, so each model must be used according to its own terms.

For this use case, a closed API might provide convenient hosted inference, but a local, open-weight approach offers something particularly valuable: the possibility of exploring nature without continuously uploading images, depending on a remote service, or paying for every inference request.

The principle behind WildLens is simple: open AI should help people spend more time experiencing the world, not more time looking at a screen.

My Agent Session

[Optional: Add a DevRelay agent session URL or embed your session using the tag format specified by the challenge.]

Prize Categories

[Add the applicable partner categories according to the official challenge rules.]

Built with 🌿 and open-source AI for a world beyond the screen.

🌿 WildLens: Explore Nature, Not Your Screen

Kothakula Navadeep — Sun, 11 Oct 2026 16:36:20 +0000

This is a submission for the "Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass" (https://dev.to/challenges/hacktoberfest-week1-2026-10-05)

What I Built

WildLens is an AI-powered nature companion designed to help people step away from their screens and explore the world around them.

The idea is simple: technology should encourage us to experience nature, not keep us glued to our devices.

With WildLens, users can:

🌱 Discover Nature: Capture a photo of a plant or bird and use AI to explore possible identifications.
🥾 Complete Outdoor Missions: Take on real-world challenges, such as identifying different leaf shapes, observing birds, or spending 15 minutes outdoors.
📖 Maintain a Nature Journal: Save discoveries, observations, and photographs to build a personal record of outdoor adventures.
📵 Explore Offline: Work toward an offline-first experience where AI inference and saved observations remain on the user's device.

WildLens is designed for nature enthusiasts, students, hikers, curious beginners, and anyone who wants to reconnect with the natural world.

The goal is to make the screen the shortest part of the experience: discover something, learn about it, and go outside to find more.

Demo

🎥 Video Demo : https://drive.google.com/file/d/1B0fO1NHYgujwa29I3714irAgs-Le1x-b/view?usp=drive_link

The project is intended to use an open-source web stack and locally executed AI inference, making it easier for other developers to inspect, modify, and extend.

How I Built It

WildLens is planned as an offline-first web application built around open AI models rather than a proprietary cloud-only AI API.

Technology stack:

React + TypeScript + Vite: For a responsive, maintainable web application.
Tailwind CSS: For a clean, mobile-friendly interface.
Transformers.js: To run compatible machine-learning models directly in the browser.
Open-weight vision model: A model such as "Xenova/clip-vit-base-patch32" can help compare an image with candidate descriptions and suggest possible matches.
ONNX Runtime: For browser-compatible inference where supported by the selected model and runtime.
IndexedDB: To store the user's nature journal and observations locally.
Progressive Web App (PWA): To support installation, caching, and offline access to supported features.

The intended workflow is straightforward:

Open WildLens and choose a nature discovery or outdoor mission.
Capture or upload an image.
Let the local AI model compare the image against candidate descriptions.
Review the suggested matches and learn more.
Save the observation locally and return to exploring.

Why Does Open Innovation Matter?

For WildLens, open innovation is about making AI more accessible, adaptable, and privacy-conscious.

Privacy by design

Reduced dependence on paid APIs

Freedom to experiment

Access beyond reliable internet

Community-driven improvement

The principle behind WildLens is simple: open AI should help people spend more time experiencing the world, not more time looking at a screen.

My Agent Session

[Optional: Add a DevRelay agent session URL or embed your session using the tag format specified by the challenge.]

Prize Categories

[Add the applicable partner categories according to the official challenge rules.]

Built with 🌿 and open-source AI for a world beyond the screen.

My hacktoberfest journey 2026

Kothakula Navadeep — Thu, 01 Oct 2026 11:15:58 +0000

🚀 Hacktoberfest 2026 — Let’s Build, Learn & Contribute! 🌐

Excited to participate in Hacktoberfest 2026! 🎉

This is a great opportunity to dive into the world of Open Source, collaborate with developers from around the world, and contribute to projects that can make a real impact.

💻 My goals for Hacktoberfest:
🔹 Make meaningful open-source contributions
🔹 Learn from real-world projects
🔹 Improve my coding and Git/GitHub skills
🔹 Collaborate with the developer community
🔹 Step outside my comfort zone and keep building 🚀

Looking forward to learning, contributing, and connecting with fellow developers throughout the month!

Let the contributions begin! 🔥

Hacktoberfest2026 #Hacktoberfest #OpenSource #GitHub #OpenSourceContribution #WebDevelopment #Programming #Developers #100Devs #LearningInPublic