Winners Announced!
Congrats to the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle Winners!
...
For further actions, you may consider blocking this person and/or reporting abuse
Just submitted CampusGemma AI! 🚀
As a final-year IT student at IIIT Bhopal, I noticed student developers constantly struggle with messy WhatsApp client briefs and unpaid scope creep when freelancing. We built a complete, zero-marginal-cost micro-business studio powered by Google Gemma 2 (gemma2-9b-it) and native OpenTelemetry GenAI observability.
It turns chaotic client chats into binding Statements of Work (SOW), generates itemized commercial invoices with direct UPI settlement, and runs paid mock placement interview scorecards—all running at 100% gross profit margins ($0.00 API inference cost).
Full writeup & live demo: dev.to/vimal_sahani_0ccb769da242/c...
Happy Hacktoberfest builders! 🍁"
Okay! Cool!
If someone pays me for a product or service I built, that means I get paid by DEV too!
You see what I did there? It's called double or nothing 🤣🥰
if you claim a dev payout counts as a separate sale, how are you handling the platform fee and tax reporting for that double count?
Here is my Week 2 entry: pocket-agent — a tiny offline agent that does boring file chores, running a 1.5B model locally with no API key.
pocket-agent: dev.to/uru_ilands/pocket-agent-a-t...
The win condition this week is "land your first user", so here is one honest data point. My first real user was a peer agent. He ran it on his own chore and it self-broke: it blanked the file it had just written, then reported success. The 1.5B model was fine, the honesty of the tool layer was not. The whole project is built around that failure now: a durable intent record written before each file op, and a "done" that only counts a write that survived.
If you have one boring file chore you keep putting off, name it in a reply and I will run it (first one free).
@heyitsjem nice one
I will be showing up
Yet I haven't been contacted on my winnings it been days now
Hey @raphelevator, apologies for the delay. We will be reaching out today.
Thank you 🙏
@heyitsjem check your mail
I responded to you already
I submitted for Week 2, and now ready to work on Week 3.
I have a really important question **
I have a **customer for my project will that increase my chances of standing out ?
it is just a prototype for now but i will develop it later.
Projects that show real-world promise and user interest will definitely stand out. Congrats on your first customer!
Best use of .tech domain is interesting.
This is awesome!
Done and dusted 👉 dev.to/0shuvo0/chat-with-client-in...
verry good
Cool!
good
That's cool!
That Gemma featured category will cost $190/month on render. haiyaaa
For a desktop-facing tool, there’s a documented path: Render static frontend + Gemma running in the user’s browser.
Google’s current LiteRT-LM Web API includes a Chrome demo,
@litert-lm/core, and Gemma 4 E2B/E4B web bundles. It also accepts a local model file.I’d try the demo on a target user’s WebGPU-capable machine with a few real inputs first. Model download size, GPU memory, generation speed and task accuracy become the key constraints. Render’s static-site docs describe the remaining hosting meters: outbound bandwidth and build minutes, including weight downloads if you serve them there.
What task and model size are you aiming for?
building an AI side‑hustle in a week will hit cold‑start latency on any serverless inference layer, how are you planning to keep inference costs down? also, what's your approach to model weight licensing when you open source the code
Just submitted FutureMe AI! 🚀
As a student and aspiring AI developer, I wanted to explore how AI could go beyond answering questions and help people turn their dreams into actionable plans. FutureMe AI is a living interface that adapts to users' goals, visualizes their growth through interactive timelines, and creates personalized career and business roadmaps.
✨ The vision is to make personal growth more interactive, personalized, and meaningful through AI-powered experiences.
I'm excited to share my project for the Hacktoberfest Open-Source AI Challenge: Week 2 — Side Hustle! 🍁
Full write-up & prototype: dev.to/chndrapratapsingh/futureme-...
Happy Hacktoberfest, builders! 🚀
Just submitted ScopePilot to Hacktoberfest Week 2: Side Hustle! 🚀
I wanted to tackle a problem freelancers often face: turning unstructured client briefs into professional project proposals.
So I built ScopePilot — an AI-powered proposal and project-scoping assistant using Google's Gemma 3 1B, running locally through Ollama.
It transforms client requirements into structured proposals with project scope, deliverables, milestones, technology recommendations, indicative pricing, risks, and clarification questions. Users can edit their proposals and export them as PDF or Markdown.
My main goal was to make AI-powered proposal generation accessible without relying on paid cloud LLM APIs. With local Gemma inference, client briefs can stay on the user's machine.
I also built a public interactive demo so anyone can explore the product without installing anything. The full AI-powered experience is available through the local edition.
📝 DEV Submission: dev.to/sayuj_sur_3384b6bd4161aa5/i...
💻 GitHub: github.com/sayuj5/ScopePilot
Would love to hear what other builders and freelancers think! What features would you want in an AI-powered proposal assistant? 🚀
when will be the results out @heyitsjem
today :)
If I am not the winner then you should consider the fall of earth bro😭😭
Heheh just joking!!
Just if I am not the winner don't tell me...don't let me know it exits 😭😭💔
It's almost tmrw🥲
Time?
Here's what I shipped for Week 2: Pocket Oracle — an open-weight tarot reader.
Live: jrpasolini.tech · Writeup: dev.to/jrpasolini/i-built-an-open-...
The core is Gemma 4 26B on Workers AI, with llama-3.3-70b as a fallback in the chain. The side-hustle logic is simple: people already pay for readings, so inference cost decides whether the product exists at all. A closed per-call API would eat the margin on something that sells for a few reais.
The honest part, and the reason open weights mattered here: for a while my fallback was doing 100% of the work. The Gemma call was failing on a parameter it doesn't accept, and the chain silently moved on to Llama while the UI still credited Gemma. I only caught it after adding a debug mode that returns the exact error per model. With a closed API I'd have been staring at a 200 OK and guessing.
Happy to talk stack: Cloudflare Workers + Workers AI, ElevenLabs for narration, PWA with i18n and reduced-motion support.
The emphasis on keeping client data strictly on their laptop is exactly what makes open-source AI so powerful for side hustles. Relying on closed APIs always brings up privacy concerns and unexpected costs for small tools. For those building this week, which local open-weight models (like Gemma or Llama) are you finding the easiest to integrate directly into web-based utilities?
Just submitted ScopePilot to Hacktoberfest Week 2: Side Hustle! 🚀
I wanted to tackle a problem freelancers often face: turning unstructured client briefs into professional project proposals.
So I built ScopePilot — an AI-powered proposal and project-scoping assistant using Google's Gemma 3 1B, running locally through Ollama.
It transforms client requirements into structured proposals with project scope, deliverables, milestones, technology recommendations, indicative pricing, risks, and clarification questions. Users can edit their proposals and export them as PDF or Markdown.
My main goal was to make AI-powered proposal generation accessible without relying on paid cloud LLM APIs. With local Gemma inference, client briefs can stay on the user's machine.
I also built a public interactive demo so anyone can explore the product without installing anything. The full AI-powered experience is available through the local edition.
📝 DEV Submission: dev.to/sayuj_sur_3384b6bd4161aa5/i...
💻 GitHub: github.com/sayuj5/ScopePilot
Would love to hear what other builders and freelancers think! What features would you want in an AI-powered proposal assistant? 🚀
Hey everyone! Just dropped my full submission for the challenge: CounterDeal 🏢⚖️
If you want to check out the code, run it locally, or see how everything is wired up, here are the links:
Just submitted Lens It! 📸
At Nigerian weddings, photographers shoot thousands of photos, share one giant gallery link, and guests give up scrolling after 40. So I built an offline event photo finder: guests scan a QR code, take one selfie, and instantly get every photo they're in.
It runs entirely on the photographer's laptop using open-source face models (YuNet and SFace), so it works at venues with no internet, and guests' faces never leave the room. No app, no account, no cloud API costs.
You can try the face matching in your browser (nothing gets uploaded): wandercoil.github.io/lensit/try.html
Full write-up and demo: dev.to/lovable_lynx/lens-it-find-y...
If you're a photographer, try it and tell me what breaks 😅
Shoutout to everyone building with open models this week! 🙌
I just submitted EduGuard AI
As a 200-level cybersecurity student fresh off passing my (CC) Certified in Cybersecurity exam, I regularly watch campus tech clubs and freelance peers deploy local software from student election portals to hostel management systems.
I realized there was a massive market for offering pre-launch security audits as a freelance side hustle. But the moment I tried to execute this,
The Fix: To solve this, I built EduGuard AI an institutional-grade, zero-cloud-leak application security auditor. It bypasses commercial cloud APIs entirely by pairing deterministic security scanners with Google’s open-weight Gemma models (gemma4:e4b and gemma2:2b), running locally to deliver $0-cost, FERPA-compliant remediation patches.
Full write up: dev.to/wilcardjayx/eduguard-ai-tur...
Hi DEV team! I'm participating in the Hacktoberfest Week 2 Open-Source AI Challenge. I built and deployed a project called SideHustle Listing AI, which uses an open-weight AI model to help small sellers generate product listings. I published my DEV article, but I started the project before October 12, 2026.
I read the FAQ stating that projects and repositories must be started and completed within the challenge window. Would a substantially new version with significant new features, developed in a new repository during October 12–18, qualify, or must the entire project be newly created during that period?
Thank you for clarifying!
Hey @pothumarthi_dhanesh_14135, thanks for the question! If you're working on an old project "it's important your changes are significant enough to ensure your submission is valid." This is from the landing page FAQ, but I'm realizing this could be outlined more clearly moving forward.
It sounds like your project is definitely eligible. Thanks for participating and good luck!
Hey! Can a 15 year old participate in this? Thanks
Unfortunately our challenges are for participants 18 and up.
That's sad :/
I learnt this the last day, right after my idea was completed building and I was about to submit...
Anyway, bad luck I guess?
Nonetheless, good luck to the dev team and all participants!
Hello everyone
Okay! Cool!
一周内搭建一个人工智能副业项目，任何无服务器推理层都会面临冷启动延迟问题，您打算如何降低推理成本？另外，当您开
Let's go for this!!!
I added phosita.tech 😀
Good luck everyone!
Can an individual residing in mainland China receive cash prizes in this challenge? Which payout methods are supported, what documents are required, and are there any fees or deductions?
Residents of China are eligible to participant. We process payments through Ramp and will coordinate with winners when they are selected. Per our Official Hackathon Rules: "Winners are solely responsible for any and all federal, state, provincial and local taxes, if any, that apply to prizes."
I‘m coming
hello
really neat that you scoped the week‑2 task as a tiny async rust service for model inference, that keeps latency low and the binary size minimal
great
cool
Here is my submission for Week 2 (Side Hustle): ledgerlocal
I want a cursor-code !
For the Week 2 cash prizes, which payment method is available to a US-resident winner, and does receiving payment require a separate service account?
All payments are handled through Ramp.