An AI-powered personal assistant accessible via WhatsApp, featuring intelligent calendar management, email composition with multi-draft workflows, and resume summarization — built on a LangChain ReAct agent architecture.
Managing schedules, emails, and document summarization across multiple apps creates daily friction. This project consolidates those tasks into a single conversational WhatsApp interface powered by an agentic LLM with tool-use capabilities.
- Designed a ReAct agent via LangChain that reasons, selects tools, and executes multi-step tasks autonomously
- Built a dual-memory system — short-term ConversationBufferMemory for context, plus long-term PostgreSQL + ChromaDB for user-specific semantic recall
- Integrated Google Calendar API for natural-language event creation, updates, and queries
- Engineered Gmail bidirectional sync with multi-draft composition, improvement loops, and numbered selection
- Powered by Groq's llama-3.1-8b-instant via a two-part architecture — Node.js WhatsApp frontend + Python FastAPI backend
FastAPI
Python
LangChain
Node.js
PostgreSQL
ChromaDB
Groq
Google Cloud
SQLAlchemy
DockerArchitecture overview — WhatsApp client connects to Node.js frontend, routing messages to Python LangChain ReAct agent backend with Groq LLM, PostgreSQL, and ChromaDB.
Calendar management — natural-language event creation and querying via Google Calendar API integration.
Email workflow — multi-draft composition with improvement loops and bidirectional Gmail sync.
Resume summarization — the agent processes and condenses professional documents into structured summaries.
