Daniel Barría
Tech Lead building toward Growth Engineering
I build full-stack systems for devtools & AI startups — backend infrastructure, AI pipelines, and the growth instrumentation that turns a site into a system teams can measure and ship on.
About
I'm a Tech Lead at Litebox, where I've spent 6 years leading full-stack web platform and growth engineering for devtools and AI companies — MotherDuck, CodSpeed, Ampersand, and ~10 other client accounts since 2023. I've built backend services end-to-end (NestJS, MongoDB, BigQuery, Spring Boot), shipped a Python/AI pipeline, and led growth instrumentation across a dozen production sites. I think of myself as an engineer rather than a programmer: I move across stacks as needed — Node, Python, Java, Go — and increasingly, I bring that experience to bear curating and reviewing AI-generated code, not just writing my own.
Experience
Tech Lead — Litebox
2020 — PresentLeading full-stack web platform engineering for devtools & AI startups. Technical Leader for MotherDuck (since 2023, the agency's largest account) and Ampersand; technical support for CodSpeed. Led ~10 client engagements since 2023 spanning frontend, backend, infra, and growth. Also run technical hiring assessments and AI-assisted performance reviews for the engineering team.
Selected Work
MotherDuck — AI Video Pipeline
Built an automated pipeline that takes MotherDuck's YouTube uploads, transcribes them with OpenAI Whisper, translates the transcripts, and publishes them straight into Strapi — all in Python. Removed a fully manual step from the content workflow.
MotherDuck — Dependency Security Campaign
Reduced security advisories from 317 to 124 (zero critical) across the monorepo with no regressions. Also built a Strapi link checker that caught 264 real broken links and cut check time from 4 minutes to 33 seconds.
Ampersand — Site Relaunch
Shipped ampersand.ai end-to-end as lead engineer on the Astro rebuild — PostHog and GTM instrumented with a pre-launch baseline, plus the final site-wide QA sweep. Backed the Phase 2 estimate with measured velocity, and got 71% of legacy URLs redirected 1:1 to the new docs.
CodSpeed — Homepage v2 & SEO
Led the Next.js App Router rebuild and owned SEO, GEO, and AEO: dynamic OG images, sitemap, robots, and structured data for AI-crawler visibility.
LiteDesigner — AI Tooling Kickoff
Kicked off an internal MCP server for Litebox's design team to capture and reuse website references, straight from Claude. Built the backend, auth, and a working 0.1 frontend with the core flow: Next.js, Neon/Drizzle, Better Auth, and an AWS worker for automated site capture. Now in active development, with the team building on top of that foundation — including code review I keep doing on incoming PRs.
imetriq — Metrics Platform (Backend-led)
Lead builder of an ad-metrics and alerting platform: a Next.js app plus an internal NestJS service querying BigQuery and MongoDB, deployed via Docker to AWS. Owned the product and its build/deploy infrastructure end-to-end.
DrLullaby — Full-Stack Across Three Languages
Full-stack + DevOps on a sleep-health platform: features across a Vue 2 frontend, a NestJS backend, and a legacy Spring Boot (Java) service — plus CI/CD to EC2 and migrating secrets to AWS Secrets Manager. One engagement, three different stacks.
Time Tracking Bot — Internal Automation
Built a Slack-based internal tool used by the whole agency: interactive forms that write time entries into Runn, with calendar integration. NestJS backend, Dockerized and deployed to AWS ECS Fargate via CI/CD.
SEO Crawler (Go) — Technical Oversight
Ongoing code review and technical judgment call on a Go-based SEO crawler service — reviewing incoming PRs, flagging edge cases, and keeping quality high on a codebase outside my day-to-day stack.
Personal Builds
Self-Hosted Infrastructure & AI Platform
A homelab running 15+ Dockerized services — Home Assistant, Immich, Jellyfin, Syncthing, AdGuard Home, qBittorrent — behind a full observability stack (Prometheus, Grafana, node_exporter, cAdvisor) and secure remote/public access via Tailscale and Cloudflare Tunnel. It also runs local AI inference: open-source LLMs served via Ollama on a dedicated GPU box (RTX 3060, 12GB VRAM), powering a self-hosted AI agent end-to-end.
AI Second Brain
A personal knowledge system: an Obsidian vault auto-populated from 40+ repos, Slack, and Notion, mirrored to a self-hosted agent via MCP. A Telegram-based AI agent — running on the same homelab, using the local Ollama models above — runs daily check-ins against it, with secure remote access via Tailscale. An end-to-end personal AI-first system, built and operated solo.
Contact
Open to growth engineering and full-stack platform leadership roles at devtools & AI startups.