You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
classSayakMitraMajumder:
role="AI Engineer · Backend Developer · Systems Builder"university="KIIT University — B.Tech, CSE (Communication Engineering)"focus= ["LLM Applications", "RAG Pipelines", "Scalable Backend Systems"]
building="AI-powered applications at the intersection of NLP and engineering"interests= ["FinTech", "Product-Based Companies", "Distributed Systems"]
languages= ["Python", "Java", "JavaScript", "C++", "SQL"]
I architect backend systems powered by AI — from RAG-based retrieval pipelines to REST APIs built for production scale. My work sits at the intersection of Language Models, system design, and clean backend engineering.
〔 Tech Stack 〕
Specialized in:LLM APIsRAG PipelinesVector DBs (Qdrant)REST API DesignPrompt Engineering
A modular multi-agent AI platform delivering personalized career preparation, code intelligence, and adaptive learning workflows through orchestrated AI pipelines.
🏗️ Architecture & Backend
Engineered a modular FastAPI backend with a shared database/repository layer powering multiple independent AI agents from a single service layer
Integrated Spring Boot middleware handling authentication, resume parsing, and backend request orchestration between frontend and AI services
Designed persistent session architecture with MongoDB-backed pipeline and state storage, including interrupted-session recovery and graceful resumption
🤖 AI Orchestration
Built a unified platform integrating three specialized agents: AI Interviewer, Code Reviewer, and Adaptive Roadmap Generator — each operating as an isolated, stateful pipeline
Implemented multi-model orchestration across OpenAI, Anthropic (Claude), and ElevenLabs APIs, routing to the appropriate model per agent's task profile
Engineered a graph-based roadmap generation pipeline with dynamic node updates, progress tracking, and version restoration support
🤖 AI Portfolio Maker
LLM-powered backend that auto-generates personal portfolio sites
Prompt-engineered generation pipeline with context-aware LLM calls
RAG architecture using vector embeddings for personalized content retrieval
REST API layer with structured output parsing and template rendering