haruan@justino — ~/portfolio — zsh online

$

Haruan Justino

Lead Software Engineer · Software Architect

I build data platforms that handle billions of records — Python, TypeScript and AWS in production at Finetune. 16+ years shipping software, most of it leading the teams that ship it.

$

  • Python
  • TypeScript
  • React
  • AWS
  • PySpark
  • Snowflake
  • PostgreSQL
  • Docker
  • MCP

$

$ cat about.md

I lead engineering on AP Classroom, College Board's platform for AP teachers and students, at Finetune (A Prometric Company). Recent work: redesigning the partitioning strategy for its largest dataset — roughly 3 billion score records and growing ~20% a year — for 3× faster retrieval, and architecting the merge of the platform's two most heavily used teacher-facing pages.

Before that I spent a year at Wiley building things fast. A medical knowledge base API in two weeks, in Python and AWS Chalice, powered the debut of Wiley's Life Sciences AI Agent — 45 press articles with a combined reach of 145M monthly unique visitors. An MCP integration on top of it, built in days, enabled Wiley's technology partnership with Anthropic. Alongside it, a large-scale data platform in Java and Spring Boot processing millions of scientific articles for LLM training and RAG.

The stack has changed a lot over 16 years — iOS, PHP, Java, now mostly Python and TypeScript. What hasn't is the part I care about: turning a new idea into something shipped, and leaving the codebase and the team better than I found them.

$ git log --author="haruan"

  1. Finetune (A Prometric Company)

    current
    Lead Software Engineer · Remote

    AP Classroom — College Board's platform for AP teachers and students.

    • Redesigned the partitioning strategy for the platform's largest dataset — ~3B score records and growing ~20% annually — on an S3 data lake queried via Redshift Spectrum, optimizing Parquet partitions for a new administrator- and coordinator-facing report at far larger scale, achieving 3× faster retrieval in testing; ran cost-estimation POCs to guide infrastructure trade-offs.
    • Architected the technical approach and led implementation of an epic merging AP Classroom's two most heavily used teacher-facing pages in React, working across legacy JavaScript and modern TypeScript, improving teacher workflows and platform performance.
    • Lead agile engineering workflows across a fully distributed remote team building a distributed system of services exposing REST and GraphQL (Graphene) APIs, balancing feature delivery with technical sustainability.
  2. Wiley

    Lead Software Engineer · Hybrid
    • Architected a large-scale data platform in Java, Spring Boot and Hexagonal Architecture, processing millions of scientific articles — extracting licensing, territory and publication metadata and computing content metrics for clients building LLM training datasets and RAG applications. Its engineering standards were adopted across multiple teams.
    • Designed the platform's DAG-based orchestration on Airflow running on Kubernetes — pipeline structure, task dependencies and triggering strategy — and wrote the Python executed by each task.
    • Delivered a medical knowledge base API in two weeks with Python, AWS Chalice, Amazon Bedrock Knowledge Bases and OpenSearch, powering the debut of Wiley's Life Sciences AI Agent at an AWS event — 45 press articles, 145M monthly unique visitors of combined reach.
    • Built a Model Context Protocol (MCP) integration on top of that API in days, enabling and solidifying Wiley's strategic technology partnership with Anthropic.
    • Led a team of consultancy-provided engineers — hiring, mentoring and performance management — through an agile SDLC enriched with Lean and XP, delivering the core data platform on schedule with the company's top performance metrics: fastest PR turnaround, optimized test durations, highest code quality. Shared the team's processes company-wide through internal workshops.
    • Set the team's standard for responsible AI-assisted development — Claude, MCP, Copilot for code review, Playwright for test automation — and shipped a behavioral code analysis tool based on Adam Tornhill's hotspot methodology to guide technical-debt prioritization.
  3. Finetune

    Senior Full Stack Engineer · Remote
    • Designed the reporting architecture for the platform's largest dataset — Redshift Spectrum external tables over Parquet on the S3 data lake, partitioned by classroom to serve student- and teacher-facing score reports.
    • Implemented core stages of the platform's layered Glue/PySpark ETL pipeline, where each stage produces inspectable, logged intermediate files, enabling recovery from any failure point without rerunning the full pipeline.
    • Optimized critical PySpark jobs, cutting execution times by nearly 50% and accelerating data availability and pipeline throughput.
    • Contributed to the Architecture Council, defining the platform's long-term architectural vision and technology roadmap.
    • Implemented collaborative assessment sharing with Python (Flask), React, PostgreSQL and REST/GraphQL APIs, significantly reducing rework in assessment creation.
    • Integrated Mock-SQS into local development, eliminating AWS infrastructure costs during development and streamlining testing.
  4. Administrate

    Staff Full Stack Engineer · Remote
    • Rebuilt multiple legacy pages on a modern React and TypeScript architecture, improving performance and maintainability.
    • Implemented a full internationalization system for automated emails, enabling multi-language templates and expanding global reach.
    • Consolidated several legacy Custom Fields implementations into a unified architecture, cutting technical debt and complexity.
    • Added robust debugging capabilities to the PHP and Python codebases, accelerating troubleshooting for the whole team.
  5. CrossKnowledge

    Senior Full Stack Developer · On-site

    Corporate learning platform.

    • Architected and led development of CK Connect, a five-service product built with Django, DRF, Vue.js and Docker — launched to positive industry reviews.
    • Designed and executed the migration of big data event logs and analytics from MongoDB to Snowflake using Redis and S3, resolving major bottlenecks.
    • Optimized the content suggestion system, reducing query response times from 30 seconds to 1.2 seconds.
    • Rebuilt local development environments for 8 projects with Docker, cutting onboarding from 2 days to 20 minutes.
    • Established Python coding standards with automated enforcement across the entire engineering department.
  6. Revenda Mais

    Lead Software Engineer · On-site

    SaaS platform for automotive dealerships.

    • Migrated hosting from a single server to a distributed Nginx and Redis architecture, increasing SLA reliability for 600+ client websites.
    • Resolved deep database bottlenecks, supporting 25% more clients on the same hardware, and modernized image-processing workflows, cutting EC2 costs by 50%.
    • Created a centralized log and monitoring system covering 20 third-party integrations, reducing new support tickets by 50%.
    • Architected the platform's CRM module with PHP, Yii2 and SendGrid.
  7. Earlier career

    iOS engineering, consulting & training
    • Blogo · iOS Engineer, 2015 — 2016 — architected the Blogo iPhone/iPad app, winner of the Apple App Store "Best New App 2016" award.
    • ENG DTP & Multimídia · iOS Consultant & Instructor, 2011 — 2015 — trained corporate development teams at Correios, Banco do Brasil and Prodam; published author ("Mac OS X Lion 10.7"); guided the team that delivered Brazil's official live election-count iOS app.

$ tree skills/

languages

  • Python
  • TypeScript · JavaScript
  • SQL
  • Java
  • PHP
  • Swift · Objective-C

frameworks

  • Django · DRF
  • FastAPI
  • React
  • Flask · AWS Chalice
  • Spring Boot
  • Vue.js
  • GraphQL (Graphene)

cloud & data

  • AWS EC2 · S3 · SQS · SNS
  • PySpark
  • Snowflake
  • Docker
  • Glue · Redshift · CloudFront
  • Airflow · Kubernetes
  • Nginx

data stores

  • PostgreSQL
  • Redis
  • MySQL
  • MongoDB

ai engineering

  • Model Context Protocol
  • RAG
  • Claude
  • Amazon Bedrock · OpenSearch
  • Copilot · Cursor
  • Playwright

practice

  • Software architecture
  • Distributed systems
  • Microservices
  • Performance optimization
  • Technical leadership
  • Mentoring · code standards

$ cat education.md highlights.md

Education

Universidade Federal do Paraná UFPR

Technologist in Systems Analysis and Development

Highlights

  • MCP integration enabling Wiley's technology partnership with Anthropic
  • Knowledge base API powering an AI launch with 145M monthly unique visitors of reach
  • Architecture Council — platform architecture vision at Finetune
  • Apple App Store Best New App 2016 — Blogo for iPhone & iPad
  • Published author — "Mac OS X Lion 10.7"

$ ./say-hello

Open to lead and architect roles working on hard platform problems. The fastest way to reach me is email.

haruanm [at] gmail [dot] com