Project Info

Vowch

Inspiration

ok, so the inspiration came from my life, itself. So my name is Akil Saji, as of right now i am working, in the intersection between, ai, marketing and development. So basically i am from a middle class background, so when i first came to the city Banglore, it was hard finding the job, India as you know, is a country with high population, and high unemployment rates, so it was hard to finding a job, even after getting a Engineering degree, so then did many blue collar jobs, like, washing dishes at Taco Bell, driving Uber, food delivery(Zomato), i mean, my thing was to, go explore life, but learnt a lot about life, too. That is the space where i discovered gig work, because, if i am facing this problem, then sure as hell, many students, many people, no matter the age, face these problems too, and yes, gig work was on rise too at that time, so that is when i decided, if one day, i get to a point where, i can do something, to solve this problem, i will... so hustled hard, learnt all, i can, in development, ai, marketing etc. So yes, basically you can say, this is my inspiration

What it does

So my app is called Vowch, where it is marketplace for gig works, but with referral, so yes, it is liek many other platforms, but here, the person, who can do the gig work has to be vouched by someone, hence, filtering the right set of people, so basically, it its more aligned towards, lets any, skilled gig work, but with a network. each and everyone in the network, is vouched by someone, already in the gig network, hence maintaining the quality and trust. So i have built a client portal as an web app, where the the person, you want the job done can post a gig work, called poster, and the person who doe sth work can login, into , and also, there is an app too, as app cannot be hosted, i have pushed the code for it. So, basically, if anyone has to join, they can scan. qr or should be referred by someone, so as a builder, i have a skill passport, where each an everyone is verified though. Skill passport, so that is the skill, and each and everyone has a skill passport number which is unique. So after sign in, anyone can pst a gig, but if a perosn has to do a gig, then he has to be vouched by someone, and only with the skill passport, that guy will be able to complete, a gig, and qr code verification is there, for everything... for verifying the network, and to identify, the correct, gig person has reached the poster, and thing like that. and if that voucher behaves inappropriately, then the Cred will go down. So, in the beginning, each and everyone will get a Cred of 500, then everything depends upon the cred and the work they do

How we built it

So, i used Codex to built, with mostly with the Codex 5.6 Terra model, the bakcend is in AWS, hosted in Vercel, and pushed though Githb for version control, then used Expo for locally hosting the app and testing,a dn also used Mapbox, for getting the map coordinates for the gig work. I had some sort of architecture in my mind, but making that into a system and product was the hardest the web app.... screen flow and everything And integrating the autonomous AI agent Sentinel, run by Llama 3.3 70B model for taking actions

Challenges we ran into

Payment and escrow, was the hardest and verification on how we can verify the gig worker, is genuine as he say he is or vouched by the guy. So it is the first time i am building an app, so the app screen flow, , it was entirely new to me, couldn't figure out half the things, but still with the help of Codex and MCP, which i connected taught me a lot and helped me

Accomplishments we're proud of

Building the entire backend, two web portal, and an app, obviously and, turning our idea into an MVP

What we learned

Verification and payment still is an edge case, it will take a lot of fine-tuning an refinement, and the vouching part which is the main feature of the product is still unexplored, if this app get traction, then we may have to do something with the referral part, because, that is like an blocker, and a filter system i have used to block out the unqualified people, and frauds.

What's next

Scaling up.............. building an elite community, for skilled market professional, who will get gig through vowch.

Vowch — trusted gig-work marketplace

Vowch is a marketplace for skilled gig work built around trust. Clients can post gigs, and workers use a Skill Passport, referrals ("vouches"), reputation, and QR verification to help make sure the right person completes the work.

This repository contains the Vowch web portals, mobile app, serverless backend, infrastructure, and deployment workflow.

Project structure

apps/client-portal/ — client-facing web portal for posting and managing gigs
apps/admin-portal/ — administrator portal
apps/mobile/ — Expo / React Native mobile app for workers
backend/ — AWS SAM serverless API and worker services
infra/ — CloudFormation bootstrap and security stacks

Built with Codex and GPT-5.6

Vowch was built during OpenAI Build Week with Codex using GPT-5.6. Codex was used as a development partner to turn the product architecture into working web, mobile, backend, and infrastructure code; iterate on user flows and interface details; help integrate AWS, Cognito, Mapbox, Expo, and Vercel; and validate and refine the implementation. The product decisions, testing, and final integration were directed by the builder.

Backend and infrastructure

The serverless backend is implemented with AWS SAM application code in backend/, CloudFormation bootstrap and security stacks in infra/, and the GitHub Actions deployment workflow in .github/workflows/.

The stack uses API Gateway, Lambda, DynamoDB, Cognito, SQS, EventBridge, WebSockets, and a private S3 upload bucket. GitHub Actions authenticates to AWS through OIDC; no AWS access key or secret key is stored in GitHub.

Deployment

Before the first GitHub Actions deployment, deploy infra/bootstrap.yaml once with an administrator AWS session and copy the resulting GitHubDeployRoleArn CloudFormation output into the AWS_DEPLOY_ROLE_ARN GitHub Actions secret. Then configure the repository secrets and variables listed in backend/deployment.md and push to main.

For local validation:

npm ci --prefix backend
npm test --prefix backend
npm run lint --prefix backend
sam validate --template-file backend/template.yaml

Analysis

Compare with all teams

View

Metric

Akil Saji
66

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

CSSIn code
HTMLIn code
JavaScriptIn code
PythonIn code
ReactIn code
TypeScriptIn code
AWSClaimed
OpenAIClaimed
VercelClaimed

6 of 9 appear in the indexed code. 3 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

1.6 MB

Source files

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.