# Raphael Bernardo **Senior Android Engineer · Applied AI & Forward Deployed Engineer** Ex-Uber (contract) · Android TV, Fire OS and multi-brand OTT streaming — Media3/ExoPlayer, DRM, native JNI/C++, and agent-driven delivery. Rio de Janeiro, Brazil (GMT-3) · Brazilian + Portuguese (EU) dual citizen Languages: Portuguese (native), English (fluent), Spanish (intermediate) Open to a new engagement; my current contract can be transitioned with notice. Offshore B2B contractor from Brazil, invoicing through my own company (CNPJ). Overlaps a full US Eastern working day. +55 (21) 98885-6697 · raphaelbgr@gmail.com github.com/raphaelbgr · linkedin.com/in/raphaelbgr --- ## Summary Senior Android engineer with cross-stack reach — Uber Maps SDK (JNI/C++), OTT / CTV streaming frameworks, and broadcast-grade DRM. Ten-plus years shipping production apps across Android, Android TV, Fire OS, webOS and Tizen — including a multi-brand OTT streaming framework and 4K DRM-protected playback (Widevine, PlayReady). --- ## Professional Experience ### Nexstar Media Group, Inc. — Senior Fire OS / Android TV Engineer **Aug 2025 – Present · Contract · Full Remote** Multi-brand OTT streaming platform framework for Fire TV and Android TV. Android/Fire OS contributor on the platform engineering team; the items below are the areas I worked in, not sole ownership of the platform. - Contributed to the architecture of an enterprise OTT streaming framework using MVVM + Clean Architecture, with a plugin-based integration system for third-party services. - Developed 5 specialised Media3 / ExoPlayer implementations for VOD, live streaming, and audio with HLS adaptive bitrate. - Implemented DRM-protected content streaming (Widevine, PlayReady) with both server-side and client-side ad insertion. - Integrated the platform's analytics and audience-measurement stack under GDPR / CCPA compliance. - Built TV-optimised device-based authentication flows. - Implemented subscription management with entitlement checking and flexible paywall integration. - Worked on remote configuration enabling zero-downtime feature rollouts across a large multi-brand flavor matrix. - Created TV-optimised UI using the Android Leanback framework with d-pad / remote-control navigation. - Refactored callback patterns to reactive event-driven architecture using Kotlin SharedFlow; resolved background-audio playback issues during navigation transitions and lifecycle-aware leaks in fragment navigation. - Integrated Firebase Crashlytics and New Relic for crash reporting and performance monitoring. - Wrote unit and UI tests using JUnit, Mockk, Robolectric, and Espresso. - Run a client-approved agent-driven delivery workflow end to end: an agent picks up the tracker issue with its acceptance criteria, works the change on an isolated branch, runs the test suite alongside agent-driven end-to-end checks against a running build, and opens a GitHub pull request carrying the issue context and test evidence. GitHub Actions runs the pipeline; a second model audits the diff against the original request before anything is reported done. **Stack:** Kotlin, Kotlin Coroutines / Flow, Media3 / ExoPlayer, Dagger 2, Retrofit2, Moshi, Glide, Android Leanback, Jetpack (Navigation, ViewModel, LiveData, StateFlow, SharedFlow, Room), Firebase (Analytics, Crashlytics, FCM), Google IMA SDK, New Relic, JUnit, Mockk, Espresso. --- ### FreeCast, Inc. — Android / Android TV / Fire OS Engineer **Dec 2024 – Jul 2025 (8 mo) · Contract · Full Remote** - Returned to the Select TV and FreeCast Watch products across Android, Android TV, Fire OS, webOS and Tizen through to the end of the engagement. - Used coding agents as a primary development method on migration and refactor work, with automated verification in CI before review. --- ### Claro TV+ — Android TV / Fire OS Engineer **Apr 2024 – Dec 2024 (8 mo) · Contract via Global Hitss · Full Remote** - Greenfield Android TV / Fire OS app for one of Brazil's largest pay-TV operators. - Implemented 4K streaming via DASH / HLS protocols with multi-DRM. - Contributed to the data layer as a shared Kotlin Multiplatform (KMP) submodule consumed by Android, webOS and Tizen clients. - Multi-module MVVM architecture; Media3 playback; analytics across Firebase + Google Analytics 4. - Applied AI-assisted development across the greenfield build — code generation and review over the Kotlin Multiplatform and multi-module layers. **Stack:** Kotlin Multiplatform (KMP), Kotlin, Jetpack (Navigation, ViewModel, LiveData, StateFlow, SharedFlow, Room, DataStore), Hilt, Media3, Coroutines / Flow, Firebase (Analytics, Crashlytics, FCM), Google Analytics 4, Picasso, Retrofit2, webOS, Tizen, JUnit, Espresso. --- ### FreeCast, Inc. — Android / Android TV / Fire OS Engineer **May 2022 – Apr 2024 (1 yr 11 mo) · Contract · Full Remote** - Led the full migration of the legacy Java codebase to Kotlin; introduced Use-Case + Clean Architecture and improved end-user UX. - Implemented HLS / DASH playback for live linear TV and VOD with Widevine DRM-protected streaming. - Maintained and developed companion apps on webOS and Tizen alongside Android TV / Fire OS. - Enabled Google Cast and integrated a paid subscription gateway. - Integrated OTA tuner platforms (HDHomeRun, AirTV, Alticast, PDAQ) with GStreamer-based media pipelines for secure live broadcast over HLS / DASH. - Owned DRM, EPG, tuner orchestration, transcoding, and multi-device streaming end-to-end. - Shipped on Google Play (Select TV Android, FreeCast Watch) and Amazon Appstore (FreeCast Watch Fire OS). - Used AI-assisted development on the Java-to-Kotlin migration and on widening test coverage. **Stack:** Kotlin, Jetpack Compose, Hilt, Coroutines / Flow, Media3, Room, DataStore, Retrofit2, Picasso, Firebase Analytics, Firebase Crashlytics, Firebase Cloud Messaging, Mixpanel, webOS, Tizen, HDHomeRun, AirTV, Alticast, PDAQ, GStreamer, Widevine DRM, JUnit, Espresso. --- ### Uber Technologies — Android Engineer, Maps SDK team **May 2021 – May 2022 (1 yr) · Contract via Matchpoint Solutions · Full Remote** - Worked on the Maps team SDK module developed with JNI / C++. - Worked in Uber's RIBs architecture inside the internal monorepo (not the public repository). - Collaborated on global-scale projects across a 10K+ engineer codebase. - Shipped to Uber Rider, Uber Eats and Uber Driver apps. **Stack:** Java, Kotlin, Groovy, C++, Python, SQL, Android SDK, NDK, JNI, RIBs, Buck, CMake, Flipper, Firebase, custom logger / analytics / crash reporter, JUnit, Espresso. --- ### FreeCast, Inc. — Android / Android TV / Fire OS Engineer **Mar 2021 – May 2021 (2 mo) · Contract · Full Remote** - Opening engagement on the Select TV Android and FreeCast Watch codebases, starting the migration from Java to Kotlin. --- ### Super Revendedores — Android Engineer (Full Remote) **Oct 2018 – Feb 2021** - Migrated the full codebase from Java to Kotlin. - Implemented online sales via payment gateway, virtual magazine showcases, and SSO link from the Android app to a partner web ordering portal for cosmetic-factory restocking. - Built scalable asynchronous order processing on AWS SQS / SNS for reliable message queuing and notification delivery. **Stack:** Kotlin, AWS SQS, AWS SNS, AWS Storage SDK, Firebase (Analytics, Cloud Messaging, Crashlytics, Remote Config, Firestore, In-App Messaging, Auth), Facebook OAuth, DbFlow, Dagger 2, MVP, Picasso, Retrofit2, JUnit, Espresso. --- ### Concrete (an Accenture Company) — Jr. Android Engineer **Dec 2016 – Oct 2018** - Built the Minha CVC travel app from its initial implementation as part of the agency team; the app passed 1M+ downloads on Google Play. - Implemented shop geolocation and store-locator guidance, increasing in-store visit volume. - Implemented a travel-package itinerary feature, reducing customer-service phone-call volume. - Implemented a travel-package virtual showcase, increasing customer flow to physical stores. - Worked on the Getnet payments app; migrated MVP to MVVM; introduced RxJava2 and LiveData. **Stack:** Java 8, Kotlin, RxJava2, LiveData, MVP → MVVM, Dagger 2, Google Maps API, Retrofit2, Picasso, Firebase (Analytics, FCM, Crashlytics, Remote Config), Jenkins CI/CD, JUnit, Espresso. --- ### PraPagar (fintech) — Jr. Android Engineer **Jun 2015 – Dec 2016** - Built the PraPagar mobile-payments app collaborating with a US-based team (MaxxPotential). - Implemented online payments via QR code and geolocation. - Backend REST work in PHP / Laravel 5. **Stack:** Android SDK, PHP, Laravel 5, REST APIs, OOP, Design Patterns, Git, Agile. --- ### IBM Brasil — Software Engineering Intern **Jun 2014 – Dec 2014 (7 mo) · Rio de Janeiro · while pursuing B.S. at Instituto Infnet** - Wrote SQL queries against IBM DB2 in support of contract and sales teams; produced reporting spreadsheets and Hyperion-based BI artefacts. - Documented internal processes via SharePoint; first professional exposure to enterprise databases, REST APIs, and Agile workflows. **Stack:** IBM DB2, SQL, IBM Hyperion, SharePoint, Python, REST APIs, Git, OOP, Agile Methodologies. --- ### FASOLTI · Fábrica de Soluções em TI — Test Developer (Internship) **Jan 2014 – May 2014 (5 mo)** - Documented and implemented automated test cases in Selenium with TestNG. **Stack:** Selenium, TestNG, REST APIs, Git, OOP, Agile. --- ## Published Applications Shipped under my own developer accounts on the App Store and Google Play. ### Archive — Home Video Player — Apple TV (tvOS) https://apps.apple.com/br/app/archive-home-video-player/id6781049173 Native SwiftUI tvOS client for a self-hosted media library: HLS playback, server-side transcoding, library browsing and resume across devices. **Stack:** Swift, SwiftUI, tvOS, AVFoundation, HLS. ### Preços Cerveja — Android — Google Play https://play.google.com/store/apps/details?id=br.com.raphaelbgr.preoscerveja Compares beer prices by volume to find the best value per millilitre. Offline-first with local persistence and in-app billing. **Stack:** Kotlin, Jetpack Compose, Room, Play Billing. ### Calculadora de Cerveja — iPhone & iPad (iOS) https://apps.apple.com/br/app/calculadora-de-cerveja/id6780996197 iOS counterpart of the price-per-volume comparison app, built natively in SwiftUI. **Stack:** Swift, SwiftUI, iOS. --- ## AI-Native Engineering Practice Infrastructure I built and run on my own machines, on my own time. Nothing below lives in a client codebase or ships in a client build. ### Multi-provider LLM orchestration A dispatch daemon routing coding work across my own seven-machine fleet, exposed to the orchestrating agent as 15 MCP tools. Cost-aware routing over 8 providers and 22 models in 6 tiers, applying rate-limit backoff, per-provider cooldowns and capability matching in order. Flat-rate providers are marked serial-only, so a parallel fan-out cannot burn a subscription's concurrency limit. Encrypted job transport keyed off the existing SSH identity — RSA-OAEP + AES-256-GCM, fresh key and nonce per message — with five-source peer discovery under an explicit precedence ladder. ### Instrumentation as a definition of done I authored a language-neutral trace-contract standard: the compile-time reference graph is the specification, and the runtime trace is reconciled against it automatically. It ships a written spec, JSON Schemas, a cross-language conformance oracle, a golden-case corpus, and an agent plugin whose hook blocks completion while the graph is stale. Gate severity is scoped — agent-authored edits block at the definition of done; human commits degrade to a lint error, never a test failure. ### Closed-loop evaluation on my own TV app My personal tvOS and Android TV media player exposes an on-device HTTP diagnostic surface — trace, state, screenshot, scripted navigation, and 149 named test routines, each declaring required parameters and returning a structured pass/fail or a machine-readable 422. An agent drives the app, captures what the screen actually shows, reads the emitted events and evaluates against the contract. Compiled out of release builds. ### Fleet operations Manifest-driven deployment across five personal machines under launchd, systemd and Task Scheduler. Each application declares its install, launch and health contract; deployment is a git push. Supervision and auto-update run on decoupled intervals so a crash restarts in seconds. Cross-machine control is RSA-signed per request. **Stack:** Python, asyncio, MCP, HTTP/WebSocket, RSA-OAEP + AES-256-GCM, mDNS/Bonjour, Tailscale, SQLite, systemd / launchd / Task Scheduler, Kotlin, Ktor, CUDA, Metal. --- ## Selected Personal Projects ### Arquive — backend for the shipped Archive tvOS app above Self-hosted personal media archive and streaming server. In-browser HLS playback with GPU-accelerated transcoding cache, IPTV (M3U + XMLTV EPG + recording), DLNA/UPnP for smart TVs, distributed GPU-fleet transcoding, FAISS-backed face recognition, local-AI captions via Ollama, plus a native Apple TV (SwiftUI) client. **Stack:** Python, React, HLS, GPU Transcoding, FAISS, InsightFace, Ollama (Qwen2.5-VL), DLNA, M3U/XMLTV, SwiftUI tvOS, JWT. ### DynamicMusicApp — github.com/raphaelbgr/dynamic-music-app-showcase Android music player with three pluggable design systems users switch at runtime — Material You (Google), Neon Wave (synthwave), and Organic Flow (botanical). One Compose codebase, three radically different visual identities — design-system theming treated as a first-class architectural concern. **Stack:** Kotlin, Jetpack Compose, Material 3, MVI, Coroutines / Flow, Hilt. --- ## Education **Instituto Infnet — Rio de Janeiro · 2013 – 2018** B.S. in Computer Engineering. Software development, mathematical modelling for software, operating-systems and computer-hardware architecture. --- ## Technical Skills **Languages:** Kotlin, Java, C++, TypeScript, JavaScript, Python, Groovy, SQL, PHP (Laravel). **Mobile / TV Platforms:** Android SDK, Android TV, Fire OS, webOS, Tizen, Android Leanback, Jetpack Compose, Compose for TV, Kotlin Multiplatform (KMP), NDK, JNI. **Streaming & Media:** OTT, CTV (Connected TV), Linear TV, VOD, Live Streaming, Media3 / ExoPlayer, HLS, DASH, H.264 / AVC, HEVC / H.265, Widevine DRM, PlayReady, Google IMA SDK, Server-Side Ad Insertion (SSAI), Client-Side Ad Insertion (CSAI), GStreamer, HDHomeRun, AirTV, Alticast, PDAQ. **Architecture:** MVVM, MVI, Unidirectional Data Flow (UDF), Clean Architecture, Use Case, Multi-module, Plugin-based integration, RIBs, MVP. **Jetpack & Reactive:** Coroutines, Flow, SharedFlow, StateFlow, LiveData, ViewModel, Navigation, Room, DataStore, Hilt, Dagger 2, RxJava2. **Cloud & Backend:** Firebase (Analytics, Crashlytics, FCM, Auth, Firestore, Remote Config, In-App Messaging), AWS (SQS, SNS, Storage), REST APIs, Retrofit2, Moshi. **Analytics:** New Relic, Mixpanel, Google Analytics 4, Firebase Analytics. **Testing & Build:** JUnit, Mockk, Robolectric, Espresso, Selenium, TestNG, Buck, Gradle, CMake, Jenkins CI/CD, Flipper. **AI & Agentic Systems:** LLM orchestration, multi-provider model routing and tier escalation, agent architectures (planner / executor / evaluator / monitor), MCP (Model Context Protocol) server and tool design, Retrieval-Augmented Generation (RAG), vector databases, knowledge graphs with community detection, prompt engineering, evaluation and scoring loops, local inference (Ollama, LM Studio, Qwen2.5-VL), faster-whisper / CTranslate2 speech-to-text, FAISS, InsightFace, Stable Diffusion / ComfyUI, Agno agents, GPU fleet scheduling and VRAM accounting (CUDA + Apple Metal). **AI-Assisted Development:** Experienced with AI coding assistants and agentic build/release tooling, applied within each client's tooling policy. **Performance & Observability:** Android Profiler, Firebase Crashlytics, New Relic, playback lifecycle and memory-leak debugging. **Compliance & Security:** GDPR, CCPA, DRM, Content Security, OAuth, SSO, d-pad focus order. **Spoken Languages:** Portuguese (native), English (fluent), Spanish (intermediate).