Bruno Leite's CV

• Email: contact@brunoleite.co
• Location: Garopaba, SC, Brazil
• Website: brunoleite.co
• LinkedIn: brunoconterleite
• GitHub: brunocleite

Summary

AI engineer who writes the code. Co‑founder, CTO and principal engineer of octoATHLETE, building a production multi‑agent LLM system — nine agents on Amazon Bedrock and OpenRouter — that turns clinicians' voice notes and sports telemetry into decision support for professional sports teams.

Twenty years of backend and cloud architecture (Java, TypeScript, Python, AWS) behind it, so the AI ships as a reliable system — versioned, observable, tenant‑isolated and cost‑controlled — rather than a demo.

Skills

AI Engineering: Agentic LLM systems, Mastra, Vercel AI SDK, Amazon Bedrock, OpenRouter, Anthropic Claude, tool calling, schema‑constrained generation (Zod), prompt engineering, LLM evals (golden sets, LLM‑as‑judge), guardrails, streaming (SSE), prompt caching and token cost control, batch inference, MCP, AI‑assisted development

Languages: TypeScript, Python, Java, JavaScript, SQL

Cloud & Serverless: AWS (Bedrock, Lambda, ECS/Fargate, DynamoDB, Aurora, S3, Cognito, Step Functions, SQS/SNS, API Gateway, IoT, X‑Ray), AWS CDK, SST, Pulumi, Docker

Data & Pipelines: PostgreSQL, Apache Iceberg, Dagster, Trino, Prisma, DynamoDB, MySQL, MongoDB

Frameworks: Node.js, Hono, Svelte/SvelteKit, Spring Boot, Spring WebFlux, Spring Data JPA, Hibernate, Vitest, Jest, Playwright

Practices: Solution Architecture, Event‑driven Architecture, Microservices, Test‑driven Development, CI/CD, Code Review, Technical Leadership

Spoken languages: English (fluent), Portuguese (native)

Experience

Co-founder & CTO / Principal AI Engineer, octoATHLETE -- Bollington, UK (Remote)

July 2024 – present

• Built and shipped nine production LLM agents and four orchestrated workflows on Mastra, from clinical voice‑note extraction and training‑load analysis to a tool‑calling readiness assistant and a safety guardrail agent.

• Personally authored ~1,550 of the platform's 2,800 commits (≈55%) across the agent layer, data lakehouse, API contracts and AWS infrastructure of an athlete health‑monitoring SaaS for professional sports teams.

• Designed a capability‑based model router across Amazon Bedrock and OpenRouter so agents request a capability rather than a model, pinning all health‑data inference to in‑region Bedrock for UK data residency.

• Kept authoritative scoring deterministic in code and LLMs to grounded narrative only — schema‑constrained output via Bedrock Converse tool calling, Zod‑validated, malformed responses retried against the failed contract.

• Authored the clinical extraction prompts enforcing verbatim evidence spans, recency and staleness gates and explicit no‑diagnosis rules, so every model assertion traces back to the source dictation.

• Built the evaluation stack — a golden‑set harness over real clinical dictation scoring capture, accuracy and false‑positive rates independently, plus LLM‑as‑judge scoring on a deterministic 15% sample of production traces.

• Ran quota‑aware batch inference on a Dagster and Apache Iceberg lakehouse, caching extractions by prompt, model and content hash and tuning concurrency to Bedrock token admission limits.

• Implemented the streaming chat layer (SSE over the AI SDK protocol) with prompt caching and per‑step token accounting, cutting stream payload ~99% and hardening it against client‑forged tenant context.

• Deploy and operate the stack on AWS with SST and Pulumi — Mastra on ECS Fargate, Lambda, Aurora, DynamoDB, Trino — and lead three engineers as architect and reviewer, holding production incidents and escaped bugs per release near zero.

Lead Software Engineer & Technical Architect (Part‑time contract), Gendo -- Remote

June 2024 – July 2025

• Established an AI‑assisted development methodology across a five‑developer team, roughly tripling commits and pull requests delivered per developer.

• Led the cross‑functional team as project leader, owning architecture, code‑quality and release standards end to end and reinforcing them through hands‑on reference implementations and mentorship.

Development Lead, FalconHR -- Remote

Aug 2023 – Mar 2024

• Architected and built a multitenant HR SaaS platform on AWS serverless, rated "very sophisticated" in an AWS Well‑Architected Review.

• Created a CI/CD pipeline deploying serverless applications together with their infrastructure in under ten minutes.

• Delivered all infrastructure as code with AWS CDK and SST (Lambda, Step Functions, Cognito, SQS, SNS), plus a DynamoDB single‑table design with streams‑based ingestion.

• Established code standards, testing methods and repository organisation, spreading them across the team through 50+ pair‑programming sessions.

Full-stack Engineer, PLS Logistics Services -- Remote

Jan 2023 – Aug 2023

• Built and extended AI/ML components that automated document revision and invoice release, replacing manual review steps in the logistics workflow.

• Introduced testing standards for the Java, Spring and PostgreSQL codebase and modernised legacy components under delivery deadlines.

Cloud Architecture Consultant, Catalyst UX (An Unosquare Company) -- Remote

Aug 2022 – Jan 2023

• Defined and enforced architecture, build, code and test standards for a Node.js, TypeScript and AWS CDK application, including a codebase‑wide refactoring to CDK standards.

• Built a shared library letting developers create on‑the‑fly deployment pipelines with AWS CodePipeline, AWS CDK and TypeScript.

• Created unit‑test standards for AWS CDK and Lambda TypeScript components, with build scripts and automated library releases via semantic‑release.

Senior Java Development Consultant (Contract), STEP -- Jersey (Remote)

Mar 2021 – Oct 2022

• Split a multi‑bank investment‑analysis application into microservices, building the GitHub Actions deployment jobs, Dockerfiles and server setup scripts behind it.

• Cut call times by optimising SQL queries and JPA implementations, and built an asynchronous SQS/S3 background processor to accelerate XLS uploads.

Senior Java Engineer (Contract), SWYP -- Remote

Sept 2021 – Mar 2022

Delivered fixes and improvements across Spring Boot microservices (Firebase, MySQL, SQS, S3) and their Docker/CircleCI deployment pipeline to Amazon ECS.

Back-end Engineer (Contract), Shoshin Works -- Remote

Mar 2021 – Sept 2021

• Built an application for NASA to predict the future of aviation, feeding multi‑source data through an NLP processor into predictive visualisations.

• Designed abstract Java components on Spring Data and MongoDB Atlas that made adding external sources such as SerpApi and the Zyte API straightforward.

Java Software Engineer, Wearsafe Labs -- Remote

Jan 2019 – May 2021

• Developed Java AWS Lambda functions processing IoT messages from personal safety devices (AWS IoT rules → Kinesis), with a critical‑section lock preventing concurrent state corruption from parallel device messages.

• Built a zero‑downtime continuous‑delivery strategy with rollback, plus an end‑to‑end test platform covering AWS IoT → Kinesis → Lambda → DB.

• Redesigned the database integration‑test architecture — tests up to five times faster with one third of the code — and eliminated duplicate message processing with ElastiCache.

Lead Software Architect (Contract), Dandelion Development -- Remote

July 2019 – Nov 2019

• Architected an ERP‑to‑CRM integration with a data warehouse on Spring Boot 2 and Spring 5 WebClient, building most of the application and leading two developers.

• Established a per‑sector test‑case structure — controllers, database, entities — reaching 75%+ overall coverage.

Java Engineer, Project44 -- Remote

Jan 2018

Built carrier integrations on the P44 API, mapping external provider models onto internal modelling with full unit and integration test coverage.

Java Lead Engineer, Aon Inpoint -- Remote

2017 – 2018

Architected a Java 8 / Spring Boot survey and benchmarking platform — REST services, multi-language surveys, an email queue, JWT security and a Docker/CircleCI pipeline.

Senior Java Software Engineer, IBM -- Porto Alegre, Brazil

2013 – 2015

Built Jakarta EE components (SOAP/REST, JPA, TDD) and cloud environments for the IBM Liquid crowdsourcing platform; automated WebSphere deployments in Python.

Senior Java Software Engineer, ADP -- Porto Alegre, Brazil

2011 – 2012

Developed security-focused Java components for ADP's portal application and contributed to its migration to a PaaS model.

Education

PUCRS (Pontifícia Universidade Católica do RS), BS in Computer Engineering -- Porto Alegre, Brazil

2005 – 2013

Open Source and Writing

• spring-boot-exception-handling — reference implementation for REST API error handling (491 stars, 275 forks), with the companion Guide to Spring Boot REST API Error Handling on the Toptal Engineering Blog.

• setup-sst — GitHub Action that provisions SST v3 (ION) with dependency caching.