LLM Architecture Explained: From Tokens to Text
2026-03-03A practical walkthrough of modern LLM architecture: tokenizer, transformer blocks, attention, training, and inference-time optimization.
I like starting from a blank slate and turning it into a finished product. From frontend and backend to deployment, I focus on building software that's practical, usable, and reliable.
Pune · Remote
Pune, Maharashtra, India · Remote
Pune, Maharashtra, India · Hybrid
Pune, Maharashtra, India · Remote

AI-powered work orchestration platform that extracts tasks from chats and emails, summarizes meetings with transcript + RAG workflows, and centralizes priorities in a dashboard with mindmap visualization and admin controls.

AI-assisted SAR drafting system for AML workflows with multi-stage LLM validation, strict compliance guardrails, role-based review, and sentence-level traceability for regulator-ready auditability.

A real-time collaborative rich-text editor built with Yjs CRDTs, enabling multiple users to edit shared documents concurrently with low-latency sync, presence awareness, and offline-first persistence.

A multi-tenant expense reimbursement platform with configurable multi-level approval workflows, OCR-based receipt scanning, and built-in currency conversion for global teams.
A practical walkthrough of modern LLM architecture: tokenizer, transformer blocks, attention, training, and inference-time optimization.
How read replicas reduce primary database load, improve read throughput, and what consistency tradeoffs to watch in production.
Languages
Frontend
Backend
Data
Cloud
Have a project in mind? Let's create something amazing.