Jafet Brito Perez
Senior Software Engineer · AI, Full-Stack & Cloud
Summary
Senior Software Engineer with over 10 years of experience architecting and deploying AI-driven search and assistant solutions. Skilled in building secure, scalable data ingestion pipelines and backend integrations using Python, TypeScript and AWS, and in implementing retrieval-augmented workflows and prompt engineering. Led the development of an AI meeting assistant and multi-agent conversational agents, significantly improving accuracy and operational efficiency.
Education
Experience
- Developed an AI meeting assistant that recorded Zoom, Teams and Meet calls as well as in-person sessions via mobile; used AWS Lambda and S3 to store audio, applied Llama 3 for note generation, drafted follow-up emails and created CRM tasks. The human-in-the-loop review significantly reduced manual transcription time and increased task-creation accuracy.
- Applied a policy-as-code compliance engine to run pre- and post-meeting checks, enforce retention and attestations, and flag restricted terms; the auditable decision logs enabled faster regulatory reviews and lowered compliance audit effort.
- Engineered conversational recommendation agents using a fine-tuned Llama 3 model with retrieval-augmented generation over Pinecone vector stores, a multi-agent debate system built with LangGraph, and prompt engineering strategies, improving recommendation relevance for knowledge-intensive tasks.
- Orchestrated the design of a decentralized MARL framework with Ray/RLlib and PettingZoo, resolving credit-assignment challenges in cooperative agents by implementing QMIX and MADDPG alongside custom reward shaping.
- Deployed CRM MCP connectors for Salesforce, Wealthbox, Redtail and Microsoft Dynamics with robust token and refresh handling, idempotent writes and conflict-resolution logic, ensuring reliable data synchronization and fewer user-edit conflicts.
- Architected and optimized high-performance RESTful APIs with FastAPI (Python) and Node.js (TypeScript), aligned with enterprise architecture standards, applying Redis caching and indexing strategies to improve data-integration throughput to 60,000 rpm and reduce API response times.
- Designed and optimized GraphQL backends, solving the N+1 query problem with DataLoader for efficient batching and caching of database queries, improving performance and scalability for complex data fetching.
- Boosted application performance with AWS Elastic Load Balancing, addressing RPS bottlenecks and mitigating DDoS attacks through automated traffic monitoring and rate limiting.
- Resolved concurrency challenges with optimistic locking and message queues such as AWS SQS, improving data integrity and supporting up to 500 concurrent users without performance degradation.
- Crafted a responsive, pixel-perfect website with React/Next.js and TypeScript, optimizing performance with React Query for efficient data fetching and caching, leading to a 25% increase in conversion rates.
- Engineered high-performance server-side rendered (SSR) applications with Next.js 13 and the App Router, significantly improving SEO visibility and navigation.
- Leveraged Capacitor to bridge web apps with native mobile features, shipping hybrid iOS and Android releases in half the time of separate native builds.
- Enhanced DevOps workflows by designing CI/CD pipelines with Jenkins, Docker and Microsoft environments, optimizing Dockerfile layers, multi-stage builds and Docker Hub / AWS ECR caching to reduce image build times.
- Championed the 2-year technical vision for Alexa Subscriptions, aligning 8 product lines and over 50 engineers; the platform powers all voice-driven subscription funnels (Alexa+, Prime, Kids+, Hey Disney!, Skills) and generates multi-million annual recurring revenue.
- Engineered an automated LLM evaluation pipeline benchmarking over 100 model/prompt variants weekly, reducing manual validation by 95% and gating Alexa+ launches across 20 KPIs.
- Standardized the fine-tuning pathway for in-house LLMs, improving 8 conversational KPIs.
- Spearheaded the design of a cross-device content recommender deployed across 6 voice and screen experiences.
- Mentored 9 engineers, leading to 6 promotions to SE2/SE3; conducted over 70 technical interview loops and trained 5 new interviewers.
- Served as Alexa on-call incident lead for 10 high-severity events, cutting MTTR by over 50% and establishing organization-wide ops reviews, design-review forums and knowledge-sharing practices.
- Eliminated 190 ms of user-perceived latency through parallelization and speculative execution.
- Formulated an automated testing and release strategy with Git-based version control and full CI/CD pipelines, replacing manual testing and release procedures.
- Designed and delivered multimodal (voice and visual) feature discovery for Echo Show (2nd Gen), improving discovery and increasing feature usage.
- Architected and launched a bulk Alexa skill enablement service, enabling skill bundling products.
- Integrated third-party payment gateways, email services and social authentication APIs through secure RESTful endpoints.
- Designed and launched voice-activated subscriptions for the Hey Disney! app with AWS Lex and Node.js, defining JSON data contracts and automating deployment with Jenkins; the voice interface enabled hands-free sign-ups.
- Led the design and development of large-scale web and mobile applications for enterprise and startup clients with React, Next.js, Node.js, Go and Python (FastAPI), PostgreSQL and Redis.
- Translated requirements into scalable architecture diagrams, GraphQL APIs and AWS deployment pipelines (ECS, Lambda, S3, CloudFront), enabling faster release cycles.
- Engineered event-driven microservices for analytics, notifications and behavior tracking with Kafka, Redis Streams and GraphQL federation gateways, increasing reliability and throughput.
- Developed real-time communication layers (WebSocket, Server-Sent Events and WebRTC) for live dashboards, chat and telemetry, significantly reducing latency.
- Unified front-end codebases with a shared React / React Native component library and monorepo tools (Nx, Turborepo), cutting development cycle time by more than 25%.
- Implemented CI/CD pipelines with GitHub Actions and Docker, automating integration tests and staging previews and enabling zero-downtime deployments.
- Integrated third-party APIs (Stripe, Firebase, Twilio, Google Maps) and built secure authentication flows with JWT, OAuth 2.0 and AWS Cognito for multi-tenant applications.
- Mentored engineers on clean code, TDD, peer review and observability, and took part in architecture reviews, hiring panels and internal governance shaping engineering standards for scalability, testing and security.
- Delivered React-based customer dashboard features, including Autoship updates, order tracking and pet profile management, in close collaboration with product and UX teams.
- Migrated legacy services to AWS by containerizing workloads and deploying them on Amazon EC2 with Terraform, with CI/CD pipelines in GitHub Actions.
- Reduced P99 latency on high-volume order and inventory APIs through distributed caching, optimized SQL queries and performance analysis during Black Friday and Cyber Monday.
- Developed operational monitoring dashboards for application performance, API errors and fulfillment delays, supporting faster troubleshooting.
- Integrated machine-learning recommendation APIs into customer-facing applications to deliver personalized product suggestions.
Skills
AI / ML: LLM development & fine-tuning, RAG, LangChain, LangGraph, Langbase, LangSmith, Hugging Face, AutoGen, CrewAI, DSPy, OpenAI API, Llama 3, PyTorch, TensorFlow, Ray, MLflow, Pinecone, multi-agent systems, prompt engineering & evaluation, Amazon SageMaker, Google Vertex AI
Languages & runtime: Python, JavaScript, TypeScript, Go, Java, C#, C++, Rust, Julia, CUDA, vLLM, Hugging Face Inference
Backend & API: REST, GraphQL, gRPC, Node.js, Flask, FastAPI, Django, Gin, Fiber, ASP.NET Core, OAuth 2.0, JWT, SSO, RBAC, ABAC, Prisma, TypeORM, SQLAlchemy, Redis, Kafka, RabbitMQ, AWS SQS, WebSockets, WebRTC, SSE, Pub/Sub, Amazon API Gateway, NGINX, Kong, Cloudflare
DevOps & cloud: AWS (ECS, Fargate, Lambda, S3, RDS, CloudWatch, EventBridge), GCP, Azure, Terraform, AWS CDK, CloudFormation, Ansible, Docker, Kubernetes, CI/CD, GitHub Actions, Jenkins, GitLab CI, Prometheus, Grafana, ELK Stack, OpenTelemetry, SLO/SLA/SLI design, VPC & networking, VPN, Zero Trust security, auto scaling, enterprise architecture, Microsoft environments
Frontend: React, Next.js, Redux, TanStack Query, SWR, SSR, SSG, SEO, Core Web Vitals (LCP, CLS), TTI, Tailwind CSS, Styled Components, MUI, Figma, Motion, D3.js, Chart.js, Webpack, Vite, Angular, Vue.js
Testing: TDD, BDD, pytest, unittest, Jest, Mocha, Vitest, xUnit, NUnit, Cypress, Selenium, Playwright, model evaluation pipelines, Pandera, k6, Locust, JMeter, OWASP ZAP
Languages: Spanish (native) · English (professional) · French (basic)