About the Project
This project is an advanced AI-driven system designed to automate and optimize the lead qualification and outreach process. It leverages Generative AI and vector similarity scoring to score leads, enrich data, and generate personalized email outreach.
The system was built to address the scalability challenges of processing over 50k leads per month, significantly reducing manual effort.
Project Goals
- To improve lead qualification accuracy using AI-driven scoring.
- To automate the enrichment of lead data from various sources.
- To generate personalized, context-aware email outreach content.
- To reduce the manual operational overhead of the sales team.
Technologies Used
- LangChain: Framework for building applications with LLMs.
- GCP Vertex AI: Used for hosting and invoking Generative AI models.
- FAISS: Vector database for efficient similarity search and scoring.
- PostgreSQL: Relational database for storing lead data and pipeline state.
- Docker & EC2: Containerization and deployment on AWS infrastructure.
- Scrapin API: Used for data collection and enrichment.
Approach
Lead Enrichment & Scoring
- Implemented a pipeline to enrich lead data using Scrapin API.
- Used FAISS for vector similarity scoring to identify high-intent leads.
- Ranked leads based on engagement potential and fit.
GenAI Outreach Generation
- Built a GenAI email generator using Vertex AI and marketing frameworks like AIDA and BAB.
- Automated the creation of personalized email drafts for top-tier leads.
Deployment & MLOps
- Deployed the entire system on AWS EC2 using Docker containers.
- Implemented GitHub Actions for CI/CD and S3 for data versioning.
Results & Insights
- 60% improvement in lead qualification accuracy.
- 80% reduction in manual outreach planning effort.
- Successfully scaled to process 50k+ leads/month.
- 45% improvement in overall lead conversion rates.