I am the Founder & Lead Engineer of Converza AI, architecting the first ConTech and Spanish-first, production-grade AI agent workspaces for heavy enterprise industries (AEC). My expertise lies at the intersection of high-performance distributed backends, custom multi-agent routing systems, and advanced hybrid retrieval infrastructure.

With a deep background in competitive programming (ICPC Finalist) and extensive software engineering experience delivering web-based generative AI UI-frameworks for firms like Bain & Company through Applaudo, I focus on building resilient, deterministic LLM systems that run at scale.

Converza AI • LinkedIn • Email

• Agentic Frameworks & Routing: Production Pydantic AI systems, multi-agent orchestrations, tool/function calling logic.
• Data & Vector Infrastructure: Hybrid Postgres RAG (pgvector + full-text), RRF fusion, Cohere reranking, multi-format parsing pipelines.
• Backend & Systems: Asynchronous Python (FastAPI, Asyncio, Pydantic), C++, WebSockets/SSE streaming, distributed systems.
• MLOps & Observability: Enterprise logging and tracing via Logfire, Sentry, PostHog, hyperparameter optimization.

• Architecture: Enterprise-grade monorepo combining a high-concurrency FastAPI streaming backend with a Next.js 14 App Router UI.
• Core Mechanisms: Engineered a production Pydantic AI agent loop that dynamically calls system-level tools (search_kb, read_kb_pdf, edit_document) based on explicit, structured input validation.
• Retrieval Pipeline: Implemented an state-of-the-art hybrid search index in pgvector utilizing full-text keyword fusion, Cohere rerank API calls, and vision-based PyMuPDF rasterization to embed scanned blueprints and graphical drawing layouts.
• Stack: FastAPI, Pydantic AI, Supabase/Postgres, pgvector, LiteLLM, Tiptap, Logfire.

• Architecture: High-throughput, asynchronous blockchain monitoring tool designed to capture real-time market discrepancies.
• Core Mechanisms: Utilizes asyncio loops and web3.py to stream the Sepolia testnet mempool, predicting real-time transaction profitability and programmatic liquidity interactions.
• Stack: Python, Web3, Asyncio, Uniswap V2, QuickNode.