Executive Summary

I am a Developer focused on building systems for complex problem-solving. My flagship project, ENTAERA, is an hybrid research agent that dynamically selects tools, maintains long term memory, and adapts its strategy to optimize for both performance and cost efficiency.

I am training for end-to-end development of AI systems, from initial API integration and debugging to building robust, multi provider frameworks that solve real world challenges.

Projects Delivered

ENTAERA

Autonomous Hybrid Research Agent

A system designed to ingest and synthesize information from multiple sources using a dynamic tool selection mechanism and vector-based memory, integrating both cloud APIs and local open-source models.

🔍 View Project Details

Key Features

• Hybrid Provider Integration: Routes requests across Azure, Google Gemini, Perplexity AI, and local models like Llama 3.1 8B.
• Vector Memory System: FAISS-powered semantic storage and retrieval using all-MiniLM-L6-v2 embeddings.
• Dynamic Tool Selection: Routes to Google Search, Arxiv, and custom functions based on task requirements.
• Self-Correction Loop: Implements rule-based feedback for improved tool selection over time.
• Cost & Rate Limit Management: Strategically utilizes cloud APIs alongside local models to reduce operational costs.
• Asynchronous Architecture: Built with Python, AsyncIO, and LangChain for concurrent, scalable task execution.

Technical Architecture

┌─────────────────┐ ┌──────────────────┐ ┌───────────────────┐ │ │ │ │ │ │ │ User Request ├────►│ Task Planning ├────►│ Tool Selection │ │ │ │ │ │ │ └─────────────────┘ └──────────────────┘ └─────────┬─────────┘ │ ▼ ┌─────────────────┐ ┌──────────────────┐ ┌───────────────────┐ │ │ │ │ │ │ │ Presentation │◄────┤ Summarization │◄────┤ Information │ │ │ │ │ │ Gathering │ └─────────────────┘ └──────────────────┘ └───────────────────┘

Proven Capabilities

• Four AI Services Connected: Reliably integrates Azure OpenAI, Google Gemini, Perplexity, and local Ollama.
• Smart Routing System: Analyzes query type and complexity to select the appropriate AI service while tracking costs.
• Solid Foundation: Built with professional logging, secure credential management, and comprehensive error handling.

Snap2Slides

AI-Powered Presentation Generator (Hackathon)

Transforms screenshots, images, and notes into presentations using AI-driven content analysis and theme generation.

🔍 View Project Details

Key Features

• Automated Content Extraction: Using Google Gemini Vision API
• AI Theme Generator: Smart color palette and layout suggestions
• Drag & Drop Interface: Intuitive content organization
• Multiple Export Formats: PDF, PPTX, HTML support

Technical Implementation

• Next.js 14 with App Router for modern UI architecture
• TypeScript with strict configuration for type safety
• Tailwind CSS for responsive, accessible design
• CI/CD Pipeline with health monitoring for production readiness

🛠️ Technology Stack