Andrew Mendez developed end-to-end agentic data interaction and vision analytics capabilities in the ai-solution-eng/ai-solution-demos repository. He built agentic SQL/RAG routing notebooks that enable direct tool-calling workflows, intelligent query routing, and secure environment management using Python, Jupyter Notebooks, and Langchain. His work included enhancements for secure NVIDIA API key handling, local embedding models, and reproducible environment setup with requirements management. Additionally, Andrew delivered a Dockerized Vision Analytics Demo app for HPE Private Cloud AI, supporting image, video, and RTSP stream analysis via a Vision Language Model, with deployment options using Docker, Helm, and Kubernetes for scalable, production-ready demonstrations.
October 2025: Vision Analytics Demo app delivered for HPE Private Cloud AI; Dockerized and Helm-deployable; enables analysis of images, videos, and live RTSP streams using a Vision Language Model (VLM); supports structured data extraction and export to external systems. This marks a concrete step toward scalable, ML-powered analytics demos in production environments.
October 2025: Vision Analytics Demo app delivered for HPE Private Cloud AI; Dockerized and Helm-deployable; enables analysis of images, videos, and live RTSP streams using a Vision Language Model (VLM); supports structured data extraction and export to external systems. This marks a concrete step toward scalable, ML-powered analytics demos in production environments.
February 2025 performance summary for ai-solution-eng/ai-solution-demos focused on delivering end-to-end agentic data interaction capabilities and reproducible experimentation environments. Key features delivered include Agentic SQL/RAG routing notebooks enabling a direct tool-calling workflow, planning/verification, and an intelligent router that selects between SQL generation and RAG based on user queries. Notebook enhancements added secure handling of NVIDIA API keys via environment variables, env-based embeddings initialization, a local embedding model, graph-based conversation history, and deployment options. Environment setup and data packaging improvements established reproducible environments with a requirements.txt and packaging changes, including file renames and a compressed database file for archiving. Impact: faster prototyping of agentic data workflows, improved security and reproducibility, and a readily deployable demo suite. Technologies/skills demonstrated: LLM tooling, RAG workflows, graph-based history, embeddings (env-based and local), environment management, and deployment integration.
February 2025 performance summary for ai-solution-eng/ai-solution-demos focused on delivering end-to-end agentic data interaction capabilities and reproducible experimentation environments. Key features delivered include Agentic SQL/RAG routing notebooks enabling a direct tool-calling workflow, planning/verification, and an intelligent router that selects between SQL generation and RAG based on user queries. Notebook enhancements added secure handling of NVIDIA API keys via environment variables, env-based embeddings initialization, a local embedding model, graph-based conversation history, and deployment options. Environment setup and data packaging improvements established reproducible environments with a requirements.txt and packaging changes, including file renames and a compressed database file for archiving. Impact: faster prototyping of agentic data workflows, improved security and reproducibility, and a readily deployable demo suite. Technologies/skills demonstrated: LLM tooling, RAG workflows, graph-based history, embeddings (env-based and local), environment management, and deployment integration.
Overview of all repositories you've contributed to across your timeline