Skip to content
View gabesanto's full-sized avatar
:shipit:
Coding all the time
:shipit:
Coding all the time

Block or report gabesanto

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
gabesanto/README.md

Gabriel Santo

Founding AI Engineer · Head of Engineering · Founder at Friendly

I design and ship production AI systems, data-heavy platforms, and product-grade automation for B2B companies.

I am most useful when a team needs to turn manual operations into software, move an AI prototype into production, or modernize a critical platform without growing engineering headcount proportionally.

Based in Brasília, Brazil — working remotely with international teams.

LinkedIn ↗ · Email ↗ · Friendly ↗

Building something around AI, automation, or platform engineering?
I take on a small number of consulting, advisory, and product-engineering engagements. Tell me what you are building or connect on LinkedIn.


Selected impact

  • Helped scale a B2B company from approximately 5 customers at join to 53 cumulative customers served, while operating as its sole engineer and technical lead.
  • Reduced a core operational workflow by 21×, from approximately 11.7 hours to 33 minutes.
  • Built an AI data workflow that has processed 293.6K rows across 96 completed runs, produced 150.6K usable leads, and saved 55.1 hours of manual cleanup.
  • Architected a 12-agent AI data-quality system scanning 2.8M+ records, reducing tracked issues by 96% and improving the internal database-health score from 78% to 95%+.
  • Led cloud and infrastructure modernization that reduced operating costs by 40%.
  • Built systems that enabled 5× operational growth and supported 900+ customer meetings in 2025.

Consulting & advisory

I work with early-stage B2B companies and data-heavy operational teams that need senior technical ownership without immediately building a full engineering organization.

Production AI systems

For teams with an AI prototype or manual workflow that needs to become reliable production software.

Typical work includes:

  • Agent architecture and workflow decomposition
  • Structured outputs, evals, observability, and prompt versioning
  • Deterministic safeguards and human-in-the-loop review
  • Reliability, retry, and failure-recovery design
  • Data pipelines, enrichment, classification, and entity resolution

Operations-to-product engineering

For companies running critical operations through spreadsheets, scripts, and disconnected tools.

I help turn those workflows into:

  • Internal platforms and customer-facing product capabilities
  • Automated and observable data pipelines
  • Provider integrations, migrations, webhooks, queues, and backfills
  • Reviewable workflows that preserve operator control

Fractional technical leadership

For founders who need someone to own both technical strategy and execution.

I can help with:

  • Architecture and technical roadmap
  • Product and infrastructure decisions
  • Cloud cost and reliability improvements
  • Engineering standards, security, and delivery
  • Hands-on implementation of the highest-risk systems

Current work

Head of Engineering & Founding AI Engineer — BuyerSight

Leading BuyerSight's transformation from spreadsheet-heavy outbound operations into an AI-powered B2B SaaS platform.

I own product and technical execution across architecture, Rails, Python, Go, Next.js, PostgreSQL, Google Cloud, DevOps, security, QA, and production AI systems.

Recent work includes:

  • Production AI-agent workflows using OpenAI, RAG, evals, structured outputs, human approval, and LangSmith observability
  • A production list-hygiene workflow that has processed 293.6K rows across 96 completed runs, producing 150.6K usable leads and 94 detailed reports
  • Hybrid company-resolution automation for a backlog of approximately 62K unlinked contacts
  • A multi-provider Unified Inbox with server-side categorization and reply workflows
  • An end-to-end Smartlead-to-Email Bison provider migration

Founder & Software Engineer — Friendly

Building privacy-first, local-first products for personal knowledge, time, and mobile productivity.

Current products include:

  • Notely — a local-first knowledge workspace with encrypted SQLite libraries, on-device semantic search, OCR, transcription, graph navigation, tasks, files, and Canvas
  • Timely — a private macOS calendar combining events, Apple Reminders, local meeting notes, and keyboard-first workflows
  • Barely — a privacy-first Android launcher built with Kotlin and Jetpack Compose

Across Friendly, I own product strategy, design, application engineering, security, release infrastructure, and cross-platform distribution.


Case studies

A sanitized breakdown of how I combine deterministic checks, retrieval, specialized agents, structured outputs, observability, confidence-based routing, and human review in production workflows.

How I helped transform spreadsheet-heavy operations into a multi-service product platform that supported growth from approximately 5 customers at join to 53 cumulative customers served.

A breakdown of the engineering decisions behind reducing a core workflow from approximately 11.7 hours to 33 minutes and lowering cloud operating costs by 40%.


Selected open-source work

A minimal, wallpaper-first Android launcher built with Kotlin and Jetpack Compose.

Focus: Android platform APIs, local fuzzy search, widgets, work and private profiles, foldables, tablets, DeX, accessibility, localization, release signing, CI, and performance benchmarking.

A cross-platform Go CLI for tagging, navigating, and operating across distributed project directories.

Focus: Go, SQLite, CLI design, local-first developer tooling, parallel operations, import/export, shell integration, automated testing, and releases.


Recognition

Built a wildfire prevention and reporting product using crowdsourcing and NASA data concepts.


Core stack

AI systems: OpenAI, Agents SDK, structured outputs, RAG, evals, LangSmith, human-in-the-loop workflows

Backend: Ruby, Rails, Python, Go, TypeScript, FastAPI, PostgreSQL, Redis

Product: Next.js, React, Electron, Kotlin, Jetpack Compose

Infrastructure: Google Cloud, Cloud Run, Cloud SQL, Terraform, Docker, Cloudflare Workers, Cloudflare R2


Work with me

Good fit

  • You have a valuable manual workflow that needs to become software
  • Your AI prototype works in demos but is not reliable in production
  • Your platform is slowing down product or operational growth
  • You need senior technical ownership before hiring a full engineering team
  • You need someone who can own strategy and still ship the implementation

I focus on production systems and measurable operational outcomes, not one-off AI demos.

Contact: Tell me what you are building · LinkedIn

Pinned Loading

  1. humanscript humanscript Public

    A modern, easy-to-use programming language designed for clarity and efficiency.

    Python 4

  2. TechTalks TechTalks Public

    A place to put some things I wanna share with the internet

  3. qmk-docker qmk-docker Public

    QMK Docker for Apple Silicon: A Dockerized QMK environment specifically designed for Apple Silicon devices. Easily build and compile custom QMK keyboard layouts without the hassle of local setup.

    Makefile 2

  4. nasaOnFire nasaOnFire Public

    Winner of Nasa Space Apps Challenge Brasilia 2019

    HTML

  5. pizzaPy pizzaPy Public

    A framework for building frontend only with Python!

    Python 8 1

  6. flutter-notion flutter-notion Public

    A Notion Like App POC of Focus in Flutter

    Dart 13 2