# Mario Aderman — AI Software Engineer > AI Software Engineer | Building intelligent systems that ship ## About Hello! I'm Mario, an AI software engineer. I'm into open-source projects and hackathons. My goal is to write maintainable, clean and understandable code. Always committed to the process. ## Skills ### AI / ML Python, PyTorch, Scikit-Learn, Transformers, LLMs (Proprietary & Open-weights), LangChain, LangGraph, Pgvector, RAG, MCP, A2A Protocol ### Backend FastAPI, FastMCP, Express, PostgreSQL, Supabase, Render ### Frontend TypeScript, React, Next.js, Gradio ### DevOps / Tools Git, GitHub Actions, Docker, Hugging Face Spaces, Claude Code, CI/CD ## Projects ### Cashy — AI Financial Advisor An AI-powered personal finance agent that answers natural language queries about a customized financial-structured database. Features multi-provider LLM support (OpenAI, Anthropic, Google, HuggingFace), specialized tool-calling, human-in-the-loop for write operations, chart generation, and a free tier. - Tags: LangGraph, LangChain, PostgreSQL, Gradio, Python - GitHub: https://github.com/MarioAderman/cashy-poc - Live Demo: https://huggingface.co/spaces/SeasonalFall84/cashy - Video Demo: https://youtu.be/Ln5vl8dd5aI ### The DETERMINATOR A generalist deep research agent that uses iterative search-and-judge loops to investigate complex questions from any domain. Built for the MCP-1st-Birthday Hackathon. Features multi-source search, MCP integration, HuggingFace OAuth, Modal sandboxed code execution and LlamaIndex RAG. I contributed the output text-to-speech task using Kokoro-82M model sandboxed on Modal GPU. - Tags: Pydantic AI, Gradio, LlamaIndex, Modal, MCP - GitHub: https://github.com/DeepCritical/GradioDemo - Live Demo: https://huggingface.co/spaces/DataQuests/DeepCritical ### Fintegra Receptionist Chatbot Multi-platform receptionist chatbot for a financial services firm. Routes user intent through a three-agent system: an intent classifier, an appointment booking agent with Cal.com integration, and a RAG-powered FAQ agent. Supports Instagram DM, WhatsApp, and Telegram with voice transcription (Whisper) and image analysis (GPT-4o). Built on n8n with Redis message batching and Supabase pgvector for knowledge retrieval. - Tags: n8n, OpenAI, Supabase, Cal.com, Redis - Video Demo: https://www.youtube.com/shorts/0mYRDSPPiOo - Instagram: https://www.instagram.com/fintegra.mx/ ## Contact - Email: mario.aderman@gmail.com - GitHub: https://github.com/MarioAderman - LinkedIn: https://linkedin.com/in/mario-aderman - X / Twitter: https://x.com/MarioAderman - Hugging Face: https://huggingface.co/SeasonalFall84 - Kaggle: https://www.kaggle.com/marioaderman/code - Upwork: https://www.upwork.com/freelancers/~01de5a92e5fb56c8c6 ## AI Integration This portfolio is AI-native. You can interact with it programmatically: ### MCP Server Connect your AI assistant (Claude Code, Cursor, VS Code, Claude Desktop) to query this portfolio: - URL: https://mcp.fintegra.solutions/mcp - CLI: claude mcp add mario-portfolio --transport http https://mcp.fintegra.solutions/mcp - JSON config: { "mcpServers": { "mario-portfolio": { "url": "https://mcp.fintegra.solutions/mcp" } } } ### A2A (Agent-to-Agent) Protocol - Agent Card: https://agent.fintegra.solutions/.well-known/agent-card.json - Endpoint: https://agent.fintegra.solutions/ ## Site - URL: https://marioaderman.com - Stack: Next.js 15, React 19, TypeScript, Tailwind CSS v4 - Languages: English, Spanish