Designing next-generation AI systems using Generative AI, RAG, and Multi-Agent Architectures

I specialize in building production-grade Generative AI systems, with a strong focus on:

• 🔹 Agentic AI workflows (LangGraph, LangChain)
• 🔹 Retrieval-Augmented Generation (RAG)
• 🔹 Multimodal AI (Text + Image + Video understanding)
• 🔹 Scalable AI pipelines using Cloud (AWS, Azure, GCP)

Currently working on enterprise-level AI systems combining LLMs, vector databases, and multi-agent orchestration.

Tech Stack: LangGraph, LangChain, Azure OpenAI, Python

• Built a multi-agent system with:
  • Query Router Agent
  • DataOps Agent
  • ITOps Agent
• Implemented dual vector databases for contextual retrieval
• Designed intelligent routing + re-ranking pipeline
• Supports PDF ingestion and semantic search

👉 Impact: Improved response accuracy and domain-specific reasoning

Tech Stack: Azure OpenAI (GPT-4o), Azure AI Search, LlamaIndex

• Extracts text + image embeddings from videos
• Performs semantic grouping of frames
• Detects code snippets using OCR (EasyOCR + Regex)
• Generates structured PDF documentation (ReportLab)

👉 Impact: Converts raw videos into searchable knowledge systems

Tech Stack: Vertex AI, BigQuery, Cloud Functions

• Built an image clustering pipeline
• Uses embedding-based similarity detection
• Automatically groups visually similar assets

👉 Impact: Enables large-scale image organization and retrieval

Tech Stack: LangChain AWS, Claude (Bedrock), Python

• AI agent to analyze logs and trace errors automatically
• Uses LLM reasoning on CloudWatch logs
• Supports root cause analysis workflows

👉 Impact: Reduces manual debugging effort in cloud environments

Tech Stack: FastAPI, LangGraph, AWS S3

• Built Agent-to-Agent communication system
• Supports:
  • Multi-file uploads
  • Stateful interaction history (stored in S3)
  • Admin monitoring dashboard
• Integrated JWT auth + role-based access

👉 Impact: Production-ready multi-agent orchestration framework

• Generative AI: GPT, Claude, Multimodal Models
• RAG Systems: Hybrid search, re-ranking, chunking strategies
• Agentic AI: LangGraph workflows, tool-using agents
• Computer Vision: OpenCV, OCR, image embeddings
• Backend: FastAPI, Flask
• Cloud: AWS, Azure AI, GCP
• Databases: MongoDB, Vector DBs, SQL

• 🔗 LinkedIn: https://www.linkedin.com/in/snehan-biswas-61429a1b4/
• 🎨 Instagram (AI Art): https://www.instagram.com/the__illusion__artist/
• 📧 Email: snehan.biswas.1@gmail.com

To build intelligent, autonomous AI systems that:

• Understand context deeply
• Scale seamlessly
• Deliver real-world impact

⭐ If you find my work interesting, feel free to explore and contribute!