Memory Over Models — AI Hackathon (HiDevs × Qdrant × Lamatic)
Memory Over Models — AI Hackathon
HiDevs × Qdrant × Lamatic × AI Collective
Online · November 2025
Build AI systems with real memory and retrieval — not chatbots.
About
Memory Over Models is an online hackathon designed to push builders beyond basic chatbot UIs. The goal is simple:
Build AI systems that remember, reason, and retrieve — like real products, not prompt wrappers.
Participants will create retrieval-first, memory-driven AI systems using Qdrant as the vector database. You’ll work on problems where memory is the core of the product — unstructured data search, personal knowledge systems, domain-specific intelligence, and more.
This hackathon is for students, engineers, founders, and AI builders who want to:
Learn how real-world AI systems are built using vectors
Work hands-on with RAG, memory architectures, and embeddings
Build something they can showcase in interviews or portfolios
Understand what separates toy AI demos from production-level workflows
Compete for prizes, credits, and exposure across top AI communities
You can use Lamatic to automate workflows, build agents, or orchestrate multi-step AI tasks.
This is not a typical hackathon — it’s a sprint to ship useful, functional AI products where vector search and memory sit at the heart of the system.
Themes
Unstructured Data RAG Challenge: Convert messy PDFs, docs, and screenshots into a powerful retrieval system using Qdrant.
AI Second Brain: Build a persistent memory OS for users — notes, chats, files, tasks stored as vectors.
Domain-Specific AI Systems: Build retrieval systems for health, finance, HR, law, travel, education, and more.
Updated Timeline
20 Nov, 10 PM IST — Registrations Open + Start Building Pick any problem under any theme and start working on your solution immediately.
25 Nov, 10 PM IST — Live Q&A
28 Nov, 12 PM IST — Submissions Open
30 Nov, 12 PM IST — Submission Deadline
30 Nov, 10 PM IST — Results Announced
Join the WhatsApp group for updates, help, and discussions:
Prizes
Top 3: ₹10,000 cash prize pool + feature on HiDevs + access to Founder Tech Network
Top 20: ₹2,00,000 credits of GCP GPU compute & Gemini 3 API Access
Top 20: Lamatic credits + certificate + community shout-out
Top 50: Perplexity Pro subscription — This is not the Airtel plan this is the full Enterprise plan.
Submission Guidelines
Deadline: 30 Nov, 12 PM IST
• Public GitHub repo
• README.md explaining concept, Qdrant usage, setup
• 1-minute demo video (Loom/YouTube/Dropbox)
• Basic comments/documentation
Rules
• Qdrant database usage is mandatory
• LLMs allowed, but pure chatbot UIs are not
• Teams of 1–4 members
• No sharing code across teams
• No pre-existing projects
• Any violation results in disqualification
Judging Criteria
• Functionality
• Originality
• User experience
• Depth of vector usage
• Real-world usefulness
Judges’ decisions are final.
Why “Memory Over Models” & Why Join
Most AI hacks are just chatbot demos. No memory, no retrieval, no real-world value.
But industry doesn’t hire “prompt engineers.”
They hire people who can build systems that remember, retrieve, and use information — that’s real AI engineering.
Models are easy. Memory is the hard part, and that’s where the real skill (and jobs) are.
This hackathon helps you:
Build actual products, not toy chatbots
Learn vector search, RAG, and memory systems used in real companies
Create portfolio projects that impress hiring managers
Get hands-on with Qdrant + optional Lamatic workflows
Win cash, credits, and even Perplexity Enterprise Pro access
Join a strong global community of AI builders
If you want to level up from “LLM user” to AI engineer, this is the place.
About the Organizers & Partners
HiDevs is building the world’s largest GenAI workforce through hands-on sessions, industry projects, and weekly AI interviews with agent-powered systems. HiDevs is hosting this hackathon to give builders the chance to create real AI systems using vector memory and retrieval.
Qdrant is a leading open-source vector database built for high-performance semantic search and AI memory infrastructure. All participating teams must use Qdrant as the core retrieval layer in their projects.
Lamatic provides workflow automation and orchestration for LLM-powered systems. Teams can optionally use Lamatic for ingestion pipelines, task flows, or intelligent automation — though it isn’t mandatory.
AI Collective is a global community of AI builders, founders, and researchers. They are the official community partner supporting participant onboarding, mentorship channels, and event amplification.