Sidhant Singh Rathore
Applied AI Engineer
I build AI systems that hold up in production: voice agents that answer while the call is still live, retrieval that stays grounded in real documents, and LLM infrastructure that fails over without anyone noticing.
Most recently I was freelancing for 
Geekashram, building their live learning platform. Before that I was an Applied AI Engineer at 
Outscale.ai on a real-time sales copilot.
I work mostly in 
TypeScript and 
Python, with 
React and 
Next.js on the front, and I build the whole path from model to product.
Reach me at or DM me on X.
Open to full-time roles and freelance projects
Experience
Projects
- On-Call Incident AgentAgentic incident response2026
- RLM PlaygroundRecursive Language Models2026
- Realtime AI Call AssistantReal-time AI Copilot2025—2026
- RIP ( Coding Harness )Agentic coding infrastructure2026
Writing
- Making a Site Legible to AI Crawlers: What I Found Rebuilding My OwnCrawlable and legible are different problems. A build report on entity graphs, llms.txt, and the positioning bug that had answer engines describing me as a vendor instead of an engineer.
- How We Cut Voice-Agent Latency to Sub-500ms: A Production ArchitectureA case study in building real-time voice agents: batching, model tiering, flush timers, and keeping the hot path off disk.
- Neural Networks: 1 to 100A complete ground-up guide to neural networks — from what a single neuron is, to how a full network learns through activation functions, backpropagation, and gradient descent.
- Retrieval-Augmented Generation (RAG): Enhancing AI with Real-Time KnowledgeExplore how Retrieval-Augmented Generation (RAG) boosts the capabilities of AI systems by combining generative models with real-time document retrieval, enabling more accurate, contextual, and up-to-date responses.