Hi, I'm Mario Aderman
AI Engineer
About Me
I'm an autodidact AI Engineer based in Mexico, passionate about making AI accessible and empowering human reasoning through open-source technology.
My work centers on building agentic systems and contributing to open-source ML tools that solve real-world problems. I believe the future of AI will be driven by collective effort, and I'm committed to being part of that movement through technical contribution and collaboration.
Education
Bachelor of Science in Mechatronics Engineering
Instituto Tecnológico y de Estudios Superiores de Monterrey (ITESM)
Querétaro, Mexico • Graduated December 2022 • GPA: 94/100
Technical Skills
Open Source Contributions
DeepCritical Project
Developer | Sept 2025 - Present
Agentic framework for scientific deep research. DeepCritical generates datasets of hypotheses, tests them systematically, runs simulations, and produces reports—all through configurable Hydra-based workflows. Originally conceived to address Alzheimer's Critical Assessment project by Hugging Science Org.
Key Contributions
- •Added LLM client support with comprehensive Pydantic validation for OpenAI-compatible model settings.PR#92
- •Resolved circular import errors for dependencies across the codebase.PR#23
- •Proposed a Ports and Adapters pattern for agentic memory and context management to benchmark all three providers (custom [hierarchical], baseline [FIFO] and external) performance.Issue#31
The DETERMINATOR Demo
Developer | Nov 2025 - Dec 2025
Lightweight app based on DeepCritical Project. Deployed on Hugging Face Spaces and submitted for MCP-1st-Birthday Hackathon hosted by Anthropic and Gradio.
Specs: Multi-modal input (audio, images and files), various options for both orchestrator and graph research modes, selection of inference providers, web search provider and audio output.
Key Contributions
- •Proposed magentic architecture for deep-orchestrator mode (inspired by Microsoft's agent framework).
- •Added audio output using text-to-speech (tts) task by refining the output report writing and deploying a sandbox on Modal's server for on-demand tts.PR#45
- •Enhanced Gradio UI design to improve app's UX.PR#45
Projects
Cashy - Financial Assistant
Oct 2025Agentic-AI assistant for personal finance management with PostgreSQL backend and Excel frontend via ODBC. Built proof-of-concept agentic workflow using Langflow with autonomous CRUD operations through LangChain components and Python code execution.
CTGAN-based Portfolio Optimization
Aug 2025Extending research on investment portfolio optimization using Conditional Tabular Generative Adversarial Networks (GANs) for synthetic scenario generation. Implementing and validating methods from Cifuentes et al. (arXiv:2302.02269). Built modular architecture for CTGAN generation, constraint optimization, and performance visualization.
Research Interests
Exploring the frontiers of AI across diverse domains, from healthcare to robotics to finance.
Medical AI
Neurology and cardiology applications to advance healthcare.
Expressive Robotics & Embodied AI
Making robotics more accessible and expressive.
AI in Quantitative Finance
Portfolio optimization and accessible financial tools.
Healthcare Companions
AI applications in medical assistance and patient care.
Digital Twins & AI Physics
Real-time digital representation of a complex system constrained by simulated environments.