Roberto Galán Senior AI Engineer

Roberto Galán, Senior AI Engineer

I build production AI agents that give teams hundreds of hours back every month.

Download resume (PDF) LinkedIn
10

Production AI platforms shipped end to end in the last 12 months

180+ h

Manual work removed every week across client teams

99.8%

Sales orders auto-loaded by an AI agent into SAP

10×

Faster ad production with retrieval-grounded generation

Systems in production

The problem, what I built, and the measured result.

Kynship · 2025 · Media operations

Autonomous video ad pipeline

ClaudeGeminiSunoWhisperXFFmpegComputer vision
30 h

Saved every week

01 · Business problem

Every video ad variant needed an editor: scripting, voice, music, captions and cuts done by hand, capping output per week.

02 · Agentic system

A multi-model pipeline that writes, scores, voices, scores music, transcribes and assembles finished edits, with vision checks on every frame range.

03 · IMPACT

The ops team runs it and editors moved from assembly to review.

Kynship · 2025 · Creative strategy

Creative analysis engine

Problem

Strategists reviewed ad performance creative by creative across accounts.

System

Meta Marketing API data joined with vision model reads of each creative, gated by evals.

120 h analyst hours saved weekly
Kynship · 2025 · Ad generation

Retrieval-grounded concept generator

Problem

Generic AI ad copy ignored who the customers actually are.

System

Generation grounded in 13k+ enriched customer profiles, with automated quality checks before anything ships.

10× faster ad production
Bytery · 2026 · Distribution & ERP

WhatsApp-native AI order desk

Voice, text, photo, XLSX & PDF intakeMeta WhatsApp bot SAP export
99.8%

Of sales orders confirmed automatically

01 · Business problem

Customers order through WhatsApp voice notes, photos of handwritten lists and PDFs. Staff retyped every line into the SAP  ERP.

02 · Agentic system

An agent parses each order, matches products with a confidence score per line, and routes only uncertain lines to a person before one-click SAP export.

03 · Enterprise impact

Order entry became exception handling. The team reviews the 0.2%, not the whole queue.

GreenGrowth CPAs · 2024 · Regulated finance

LLM-assisted audit back office

Problem

Audits and reporting ran on manual document review and spreadsheet handoffs.

System

LLM-assisted audit workflows plus QuickBooks to AWS to Power BI data pipelines.

30 hper audit
16 hweekly reporting
Hakkasan Group · 2010–2020 · Hospitality

Ticketing & venue platform

Problem

High third-party platform ticketing fees across Las Vegas venues.

System

e-commerce, ticketing and venue management in house.

$M saved in ticketing fees

27 years of shipping

2025 – now
Kynship
San Diego, CA · Remote

Senior AI Engineer · Agentic Systems

Own the AI roadmap end to end, partnering with leadership. Shipped 10 eval-gated, tool-calling production platforms in a year, including the video pipeline, creative analysis engine and concept generator above.

2024 – 2025
GreenGrowth CPAs
Los Angeles, CA · Remote

Senior Automation Engineer

LLM-assisted audit back office in a regulated accounting domain: 30 hours saved per audit. QuickBooks to AWS to Power BI pipelines: 16 hours saved weekly. AI voice outreach converting 1% of cold calls.

2013 – now
Bytery LLC
Buenos Aires, Argentina

Founder & Engineering Lead

Founded and scaled a full-stack studio from zero to profitability; technical lead on every engagement. Latest: the WhatsApp-native AI order desk with one-click SAP export.

2010 – 2020
Hakkasan Group
Las Vegas, NV · Remote

Lead Full Stack Engineer

Led engineering for e-commerce across Las Vegas and New York hospitality properties; owned ticketing and venue management architecture, saving millions in ticketing fees.

1999 – 2010
Investars.com
New York, NY · Remote

Senior Advisor, early team

Dot-com startup tracking Wall Street analyst recommendations: charting and financial data delivery systems. Successful exit.

What I build with

AI / LLM Ops

Agents

Tool calling, MCP, evals & guardrails, retrieval grounding, RAG, Proprietary agent orchestration harness.

Models

Claude (Fable, Opus, Sonnet, Haiku) Google Gemini, Anthropic, OpenAI and Groq APIs, Open source local models.

Generative media

Seedance, Nano Banana, Suno, ElevenLabs, WhisperX, OpenCV, FFmpeg

Core Architecture

Backend

Python, Flask, REST, SQLAlchemy, PostgreSQL, SQLite, Pandas

Frontend

TypeScript, React, Node.js, AlpineJS, TailwindCSS, Chart.js

Data

ETL pipelines, scraping at scale, vector DBs, embeddings, GraphQL, schema design

Infrastructure

Agent orchestration

Claude Code swarms, Codex, Opencode, custom harnesses

Cloud & ops

AWS (EC2, S3, RDS), Azure, Docker, Nginx, PM2, Linux, CI/CD, OAuth 2.0, observability

Local inference

Ollama, LM Studio, edge devices, Whisper

Have an AI initiative that needs to reach production?

Open to senior and lead AI engineering roles with enterprise teams and funded startups solving interesting problems with AI.

hi@robertogalan.com LinkedIn Resume (PDF)