Project · CIC x AWS GenAI Hackathon
LinkedOut — weaponized corporate satire powered by AWS
LinkedOut is a satire of modern LinkedIn culture, built during the CIC GenAI Hackathon on AWS. You type something painfully ordinary like "I opened my eyes today" and the app turns it into an over-engineered, AI-generated humble-brag post drenched in corporate jargon.
Under the jokes, it's a real serverless stack: a Streamlit frontend talking to AWS Lambda, Bedrock models, and S3 for storage — wired together with IAM, CloudWatch, and a fully managed, zero-servers backend. It's what happens when actual cloud engineering is used to roast performative productivity.
User prompt
"I opened my eyes today."
LinkedOut post
Thrilled to share that I kicked off my morning by successfully executing a full eyelid activation sprint 👁️. This small but meaningful "vision boot sequence" reminded me that growth isn't about massive milestones — it's about choosing to show up, one blink at a time.
Grateful for the opportunity to collaborate with my circadian rhythm, nervous system, and a world-class pillow that believed in my potential even when I was offline. Onwards and upwards. 🚀 #mindset #growth #eyelidLeadership
The interface is intentionally minimal: you type something boring, press a single button, and get a catastrophically over-celebratory LinkedOut post.
Prompts are engineered to maximize buzzwords, inspirational framing, and faux vulnerability while staying just barely plausible as a real post.
S3 for storage, Lambda for compute, and Bedrock for inference. No servers to patch, just events, functions, and a lot of sarcasm.
IAM roles keep Bedrock, Lambda, and S3 tightly scoped. CloudWatch tracks every generation so debugging cringe is actually observability.
We used Streamlit for rapid iteration at the hackathon: single-page app, instant reloads, and zero JS build pipeline to babysit.
Soft constraints on post length and frequency keep the app fun demo material rather than a firehose of infinite corporate monologues.
Architecture at a glance
- Streamlit frontend: single-page UI where users drop in mundane life events and view the generated LinkedOut post side-by-side.
- API via AWS Lambda: Streamlit calls a lightweight Lambda function that builds the Bedrock prompt, forwards the request, and logs metadata.
- Bedrock models: used through the Bedrock runtime client with carefully tuned system + user prompts for that cursed "LinkedIn thought leader" tone.
- S3 storage: keeps example posts and hackathon artifacts; architecture is ready to persist user slop if we ever want a timeline view.
- IAM + CloudWatch: least-privilege roles for Bedrock + S3 access, and request logs streamed into CloudWatch for debugging and post-mortems.
// Request flow
Streamlit ↦ Lambda ↦ Bedrock ↦ Lambda ↦ Streamlit
// Side-channel: Lambda ↦ CloudWatch logs, optional S3 dump
From annoyance to concept
The idea came straight from our own feeds: endless AI-written posts about trivial events. LinkedOut distilled that frustration into a focused, shippable joke.
Serious stack, unserious tone
Even though the output is comedic, the stack is production-relevant: Bedrock integration, IAM hardening, observability, and a clean serverless boundary.
Shipping in 24 hours
The hackathon constraint forced a tight scope: one core interaction, one polished flow, and a stack we could fully explain to judges in under five minutes.
Team
Built by a small crew of LinkedIn survivors who turned doomscrolling into a hackathon brief.
Arsh
Teammate 1
Click to view their LinkedIn profile and the other side of the joke: real experience behind the satire.
Kush (me)
Teammate 2
Click to view their LinkedIn profile and the other side of the joke: real experience behind the satire.
Sarim
Teammate 3
Click to view their LinkedIn profile and the other side of the joke: real experience behind the satire.
Sayuj
Teammate 4
Click to view their LinkedIn profile and the other side of the joke: real experience behind the satire.