Naveen G

About Naveen G

As an AI/ML Engineer and Data Scientist, I build scalable intelligent systems that…

Experience

1. Data Scientist

   Revscale Ai

   Sep 2024 — Present · Fort Lauderdale, FL, US

   Collaborated with cross-functional teams to design, develop, and deploy ML models and workflows supporting lead generation and multi-franchise business analysis. Integrated small language models (SLMs) for inbound/outbound agents, autoreplies, sentiment analysis, and intent classification using RAG techniques, data lakes, and external APIs- Fine-tuned LLAMA-3 8B SLMs using SFT, LoRA, and DPO for domain-specific agents, achieving 25% increase in response personalization and 30% higher client engagement across 50+ clients- Built 4 core AI Agents: Custom Inbound Agent for lead classification/conversion, Prospector Outbound Agent for automated discovery/outreach, FDD Agent for demographic targeting, and Auto Responder for multi-channel engagement- Developed multi-AI agentic workflows using LangChain and Airflow for lead gathering, extraction, validation, and classification across 5+ social platforms, improving efficiency by 40%- Streamlined outreach sequences and lead scoring with Crew AI and Airflow, creating DataLake pipelines resulting in 50% conversion increase and processing 500+ daily leads- Trained multi-class sentiment analysis and multi-label intent classification models using Microsoft DeBERTa with hyperparameter optimization, improving accuracy by 18%- Constructed Data Lake managing 100+ TB data using Azure Blob Storage and CosmosDB, developed preprocessing pipelines with Azure Function Apps optimized for multi-threading, reducing processing time by 30%- Built RAG-based text analysis models and FDD bots using Pinecone, improving data extraction accuracy by 20%- Orchestrated CI/CD pipelines with GitHub Actions, deployed on GPU-accelerated Azure VMs, integrated Kafka/Event Hubs/S3, applied QAT with Weights & Biases maintaining 95% performance