Jairo Espinosa

Jairo Espinosa Salazar Jr

Senior Full-Stack Engineer

Results-driven Senior Full-Stack Engineer with 7+ years of experience building high-traffic web applications in React, TypeScript, and Node.js. I lead engineering teams with a passion for mentorship, excellence, and experiment-driven product delivery.

My recent work spans software engineering, AI training and evaluation, and full-stack projects. I enjoy building creative, efficient solutions and helping teams deliver useful products.

When I'm not coding, I like to golf, play video games, and cook.

CCTV

A full-stack video application that turns long-form videos into short-form clips.

• Built a TypeScript backend with Node.js, Fastify, and pg-boss, plus a React frontend.
• Used OpenAI to transcribe videos and select strong moments, then cut, reframed, and captioned clips with FFmpeg.
• Moved paid AI calls into cached background jobs so retries and repeat uploads do not pay for the same work twice.
• Designed blind head-to-head voting between clip variants to learn which editing styles perform best.

• TypeScript
• Node.js
• Fastify
• PostgreSQL
• pg-boss
• React
• OpenAI
• FFmpeg

tri-review

An open-source GitHub Action and CLI for reviewing pull requests with multiple AI models.

• Sends a pull request to OpenAI, Anthropic, and Google models in parallel.
• Synthesizes findings flagged by two or more models separately from single-model guesses.
• Used in CI across personal repositories.

• Python
• LangGraph
• GitHub Actions

League of Legends Match Analyzer

A full-stack match-analysis application with a Python backend and a Next.js 16 frontend.

• Built the FastAPI and SQLModel backend with ARQ jobs and a Next.js 16/TypeScript frontend.
• Implemented Redis-backed sliding-window rate limits for Riot API app and method tiers, with reactive 429 backoff.
• Designed hybrid PostgreSQL storage with relational tables, JSONB, and pgvector embeddings, plus ARQ background processing.
• Built a RAG-enhanced LLM analysis pipeline that scores in-game decisions against population baselines for personalized improvement insights.

• Python
• FastAPI
• Next.js
• TypeScript
• Riot API
• OpenAI
• PostgreSQL
• Redis

Transaction Aggregator

A full-stack application for searching transactions from connected credit cards.

• Built a Python/FastAPI and SQLAlchemy backend with a React/TypeScript frontend, connecting credit cards from four banks through the Plaid API.
• Stored transactions in local SQLite with full-text search, giving instant results that continue to work offline without bank requests.
• Designed background sync to avoid duplicate bank requests from double-clicks, retries, or repeated webhooks without losing updates that arrive mid-sync.
• Secured bank access tokens with AES-256 encryption, alongside hashed passwords, sessions, and CSRF protection.

• Python
• FastAPI
• SQLAlchemy
• SQLite
• React
• TypeScript
• Plaid API

What I work with.