* uHavoc – Chaos Testing Platform (Havoc Scenarios)

Built and owned Havoc Scenarios, productizing RPC-level fault injection (abort, timeout, latency) with rich selectors (caller, callee, service, tier, transport, direction, etc.) to enable scalable resilience testing across Uber.

Replaced hardcoded defaults with database-backed User & Published Scenarios, simplifying CI integration and cross-team adoption.

Powering 180K+ automated chaos tests across 47 critical mobile flows, identifying 23 resilience risks and enabling AI-driven root cause analysis (Link to paper in Publications section below).

* Multi-Agent AI – Havoc Orchestrator

Designed and implemented a multi-agent AI system that translates natural language prompts into end-to-end fault injection workflows.

Built LangChain tools integrated with havoc-simulator, MCP servers, Whober, and SLATE to programmatically create and activate fault scenarios.

Reduced setup complexity for advanced use cases (e.g., mobile admin production traffic testing), lowering the barrier to large-scale resilience validation.

* Trace + GenAI Pipeline – Custom Retry Detection

Built an org-wide pipeline to detect unsafe custom retries via Jaeger trace pattern analysis.

Leveraged GenAI (AI Agent) to validate retry logic directly in code, extract evidence, and classify findings for migration or feature gaps.

Automated remediation by generating actionable Jira tickets, strengthening dependency safety and standardizing retry practices across services.

A: Menu Seller App - Developed an SDK and web application in React with a Rails backend,

enabling seamless menu imports from various food delivery sources. Transformed user-uploaded restaurant menus, whether through URLs or photos, into easily transcribable data.

B: Image Mapping Project - Utilized AI to locate images for thousands of products listed in a CSV file. Extracted product information and images from manufacturer websites, delivering a comprehensive solution for product image retrieval.

C: Retail Seller App - Engineered a full-stack SDK facilitating the transfer and import of retail products across different platforms, akin to the Plaid model for retail. Implemented OAuth authentication for secure data exchange.

Facebook feature where users can save videos, images, post into collections. User can also create different kind of collections and share with their friends

User in Instagram or Facebook from account center can turn on Shared Shopping History (SSH) which means the wishlist from IG and FB will be synced and user will be able to see the same data in both apps. So if user save an item in wishlist in IG, user should be able to see that item in FB. The same thing applies to Recently Viewed items in Shops.

Working in a big migration from native into a internal tool.

Working in the new FB Cart design. I worked in the merchant cart in terms of functionality and promotions/discounts.

User can watch different kind of videos in FB app. there are custom notifications when a creator or friend post a new video. So we send notifications to user related to interested videos.

Create new web features for an American company using Ruby on Rails as the backend framework, Postgresql as the database, and Jquery for the frontend.

Develop a platform where the students can search internships in faurecia and apply. The recruiters can upload vacancies and see which students have applied. Made in Ruby on Rails. Hosted in Heroku. www.talentofaurecia.com.mx