Architecting
Intelligence.
AI Infrastructure Engineer
I build advanced AI systems and deploy them at scale — specializing in RAG, Agentic Workflows, and Ollama, with deep experience designing on-premise AI servers from scratch.
How are you?
I'm Bhavya Dave
AI engineer building intelligent systems from the ground up — RAG pipelines, agentic workflows, on-premise LLM infrastructure, and cloud-native deployments on AWS & GCP.
Journey & Experience
Work Experience
AI Software Developer (Research and Development)
Protocase
- •Architected a RAG system using on-premise LLMs (Ollama) enabling queries across 10,000+ technical documents.
- •Developed local agentic workflows (CrewAI, Ollama) and deployed MCP servers for autonomous data processing.
- •Designed AI legal pipelines via YOLO v8 and EasyOCR on GPU-accelerated Docker containers.
- •Engineered ML pipelines for behavioral feature prediction using Random Forest, XGBoost, and LightGBM.
Software Developer Co-op (Research & Development)
Protocase
- •Established Kubernetes infrastructure from scratch to optimize Ollama LLM server response times.
- •Designed a predictive AI model to forecast Non-Conformance Reports (NCR), reducing quality control costs.
Data Science Intern
Rapidops Inc.
- •Developed a computer vision-based attendance system, improving accuracy by 32% using deep learning.
- •Engineered a 256-class dataset using advanced augmentation, securing additional stakeholder funding.
Education
Master of Applied Computer Science
Dalhousie University
Graduate Certificate in Cloud Data Analytics
Bachelor of Technology in Computer Science
Birla Vishwakarma Mahavidyalaya
Featured Projects
Built at Protocase during R&D roles
Agentic RAG System
Architected a private RAG system for 10k+ documents using on-premise LLMs (Ollama). Implemented keyword-based metadata scoring for ranked semantic retrieval, enabling natural language queries across internal technical documents.
AI Legal Document Pipeline
Automated clause extraction and compliance verification via YOLO v8 and EasyOCR on GPU-accelerated Docker containers.
Kubernetes LLM Cluster
Established K8s infrastructure from scratch to optimize Ollama LLM server response times for internal AI manufacturing tools.
Independent research & side projects
CO2 & Renewable Energy Prediction
Mar 2025 – Apr 2025
Hybrid time-series forecasting model using Prophet and Random Forest. Processed large-scale environmental datasets with Recursive Feature Elimination (RFE).
Technical Stack
AI/ML Core
DevOps & Cloud
Data & Languages
Publications
Gender Recognition and Age Detection Using Human Facial Features
Published in International Journal of Scientific Research in Engineering and Management (IJSREM)