Back

Pulakit Bararia

Founder Snippetz Lab...ย โ€ขย 6m

Iโ€™m building an AI-first architectural engine that converts natural language prompts into parametric, code-compliant floor plans. It's powered by a fine-tuned Mistral LLM orchestrating a multi-model architecture โ€” spatial parsing, geometric reasoning, layout synthesis, and real-time customization all run through independent agents. Iโ€™ve stacked transformer-based reasoning with procedural generation and constraint solvers to make outputs structurally valid and design-aware. To be honest, itโ€™s about 20% creative thinking, 60% pretrained AI models, and 10% my own code, architectural logic, and a bunch of trial-and-error hacks that pull it all together. Fully modular. Editable in real time. Outputs usable by both architects and non-architects. The goal? Let anyone design a liveable space โ€” without touching CAD. coming in a few months

8 Replies
19
31
Replies (8)

More like this

Recommendations from Medial

Harsha Vardhan

Empowering Architect...ย โ€ขย 10m

building a platform specially for architects and designers, where they can find internships, freelance projects, projects etc. this platform is not only for individuals but also for design firms are architectural firms. architects and designers can a

See More
Reply
4

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 8d

Real Agentic AI is a multi-layered system, where each layer solves a specific challenge from reasoning to compliance: 1. LLM (Core Reasoning) โ€“ Handles language understanding and generation. Alone, not enterprise-ready. 2. RAG (Retrieval Layer) โ€“ G

See More
Reply
1
Image Description

Nishant Gupta

Too lazy to think of...ย โ€ขย 10m

Iโ€™m building an exclusive luxury real estate brand that creates custom-designed homes based on the clientโ€™s personality. Each property will be a piece of art, emotionally tailored through deep psychological insights and architectural creativity. Iโ€™m

See More
1 Reply
4

AI Engineer

AI Deep Explorer | f...ย โ€ขย 9m

LLM Post-Training: A Deep Dive into Reasoning LLMs This survey paper provides an in-depth examination of post-training methodologies in Large Language Models (LLMs) focusing on improving reasoning capabilities. While LLMs achieve strong performance

See More
Reply
2

Abhinav kumar

First step was but d...ย โ€ขย 6d

In 2024, companies wasted over $50 BILLION on AI implementations that failed in production. Here's why 85% of Fortune 500 companies have blocked ChatGPT from their systemsโ€”and what you can do about it. ๐Ÿšจ THE PROBLEM: AI hallucination is costing ente

See More
1 Reply
1
Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 1m

8 common LLM types used in modern agent systems. 1) GPT (Generative Pretrained Transformer) Core model for many agents, strong in language understanding, generation, and instruction following. 2) MoE (Mixture of Experts) Routes tasks to specialized

See More
1 Reply
6
Image Description

Alpixn Technologies Private Limited

๐Ÿ”น Machine Learning ...ย โ€ขย 10m

Investor Proposal for Vastu Software Executive Summary: Our Vastu Software is an innovative technology-driven solution designed to assist individuals, architects, real estate developers, and interior designers in integrating Vastu Shastra principle

See More
1 Reply
3
10
Image Description
Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 1m

Many still see Google as โ€œjustโ€ a search engine-but thatโ€™s missing the bigger picture. In reality, Google is quietly building one of the most comprehensive AI ecosystems in the world, spanning video, images, models, agents, and code. This isnโ€™t a co

See More
2 Replies
1
6

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 1m

ChatGPT vs Claude vs Gemini. I've explained the best use cases of each. โœ… Which one is best for you? 1. ๐—ฃ๐—ถ๐—ฐ๐—ธ ๐—–๐—ต๐—ฎ๐˜๐—š๐—ฃ๐—ง ๐˜„๐—ต๐—ฒ๐—ป you need quick drafts, conversational edits, code help, or want to build a custom persona. Itโ€™s great for ever

See More
Reply
1
12
Image Description
Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 5m

In a few years, we went from: GPT-2 struggling with simple text โ†’ GPT-5 reasoning, planning, coding, and remembering context. Hereโ€™s the journey in a nutshell: ๐Ÿ‘‰ GPT-2 (2019): Simple text, lots of errors. ๐Ÿ‘‰ GPT-3 (2020): Huge leap (175B params), l

See More
2 Replies
2
11

Download the medial app to read full posts, comements and news.