Alejandro Gómez

Invisor finds and analyses properties based on an investor’s budget, location and investment strategy.

I develop the product and backend. I used Jev to label 50,000 properties in the database and prepare a feature dataset for CatBoost. I train and evaluate valuation models, compare their predictions with LLMs, and serve photo analysis with DINOv2 and ONNX.

I separate fact extraction from calculation and explanation. I evaluate models on properties and markets excluded from training and compare them with simple methods before assuming that more complexity improves the result.

A desktop app for chatting with local or hosted models and using MCP tools.

We developed it as a team. I added platform login with OAuth 2.1 and PKCE, its connection to Electron’s main process, and tracking of the model used for each response.

Users can connect their own providers or sign in with a platform account. The app records which model answered each message so its usage can be queried.

Miura builds custom learning platforms. Our client work includes Big School, Ignis Formación and Endoaula Conecta.

I co-founded it with Octavian Kneupper. I handle the backend and AI integrations. At Big School I worked on chat, sessions, permissions and releasing content according to the course calendar.

Content access depends on the student, the programme they are enrolled in and the date. The AI chat follows the same access rules.

I use Hermes Agent for Invisor’s work outside the code: sales, meetings, user analysis and organisation.

I connect it to the CRM, transcripts, Notion and Invisor data. It uses Obsidian as persistent memory: a daily cron records everything it has done and links the notes. It also reads Instagram and TikTok statistics.

Each week it checks where users leave onboarding, who buys and what they do afterwards. It gives me a report that I use to decide what to change in Invisor.