Improving

Senior Software Engineer · Remote

Full-time contractor at Improving, embedded with client teams: first Diligent, on backend and React, and since July 2024 Cinemark, on the digital ticketing platform.

• 
  CinemarkJul 2024 — Present

  Digital channel for the cinema chain: showtimes, seat selection and ticket purchase. NestJS microservices over Azure Event Hubs, React on the frontend and deployment through Azure DevOps.

• 
  DiligentSep 2023 — Jul 2024

  Corporate governance and compliance platform: backend services and React product interfaces, end to end.

• Cinemark’s backend is NestJS microservices with PostgreSQL and Redis, using Azure Event Hubs for the asynchronous flows between services. The frontend is React and pipelines run through Azure DevOps.
• I maintain the payment gateway, box-office and SendGrid integrations. They belong to third parties and cannot be changed at will, so the work is in isolating their failures and their latency.
• Load concentrates on premiere days, so the real measure of the work is whether checkout holds up through the spike without degrading.
• At Diligent the work ran end to end on the same platform: the backend services and the React product interfaces.

Randstad Digital · Torc

Lead Engineer · Remote

Leading technical architecture and delivery of scalable backend solutions for enterprise clients: Python microservices (Django/FastAPI), Rails services and CI/CD pipelines with IaC.

• 
  MoverisDec 2025 — Present

  Anti-fraud and digital identity: real-time human liveness APIs and identification of AI-generated content and video.

• Leading engineering efforts for enterprise client engagements through Randstad Digital's Torc platform. Responsible for technical architecture, team coordination, and delivery of scalable backend solutions.
• Designing and building Python-based microservices and API layers using Django and FastAPI, with emphasis on modularity, observability, and tenant-aware service patterns.
• Implementing Ruby on Rails services for internal tooling, admin dashboards, and integration middleware connecting client systems with third-party APIs and data providers.
• Developing React-based front-end interfaces for project management and delivery tracking, enabling real-time visibility into engineering progress and resource allocation.
• Establishing CI/CD pipelines, IaC patterns (Terraform/Pulumi), and deployment strategies to support rapid iteration across multiple client environments.

Fuse

Senior Backend Engineer · LATAM · Remote

Built the backend architecture of a payments fintech: onboarding, KYC with banking integrations, payment queues (Celery/Redis) and an event-driven architecture with Kafka.

• Supported the technical foundation of an early-stage fintech focused on digital payments and alternative financial services across the USA and LATAM during a short-term freelance engagement.
• Designed and implemented the core backend architecture using Python (Django/FastAPI), prioritizing scalability, modularity, and tenant-aware service design from day one. Built asynchronous task runners with Celery and Redis for payment processing queues.
• Developed initial services for user onboarding, identity verification (KYC), and virtual account creation by integrating third-party APIs such as banking partners and verification providers. Implemented the KYC orchestration flow in Python with retry logic and webhook callbacks.
• Built a React-based internal admin panel for compliance teams to review KYC submissions, flag suspicious accounts, and manage approval workflows in real time.
• Established an event-driven architecture using Kafka to power asynchronous flows for notifications, transaction reconciliation, and compliance operations.
• Defined reproducible CI/CD workflows and managed cloud infrastructure using Terraform and Pulumi to streamline deployments in staging and production environments.

Revelo

Senior Software Engineer · LATAM · Remote

Through Revelo, dedicated engineer for StyleSeat: built and operated the agent workflows that take repetitive work off professionals, plus the booking features and the Braze and Segment integrations.

• 
  StyleSeatJan 2024 — Jul 2026

  Beauty and wellness marketplace: agent workflows in production for professionals, with their tools and guardrails, plus the booking features and the Braze and Segment integrations.

• Engaged through Revelo as a dedicated senior engineer for StyleSeat, the beauty and wellness marketplace, working embedded with their product and engineering teams.
• Built the agent-driven workflows for the professionals. A stylist loses hours every day confirming appointments, chasing no-shows and rescheduling, and that time is unpaid. The agent reads the calendar and booking state, drafts client messages and applies booking changes inside the limits the professional sets.
• What was mine end to end were the tools the model can call, their schemas, and the orchestration around them: retries, timeouts and a deterministic fallback for when the model is down.
• The guardrails mattered more than the model. Availability always comes from a real call to the scheduling service and never from model text, irreversible actions require confirmation, and every action is logged with what the agent saw and what it called.
• I did not just build them, I ran them: workflows were versioned, so a prompt or tool change shipped as a tracked, reversible version, was validated in staging against real booking scenarios, and access was separated by role across product, engineering and support.
• Managed the caching layers and the resources the agents depend on: what gets cached and for how long, rate limits and token budgets per workflow, and quotas that keep one workflow from starving another during peak booking hours.
• Integrated Braze for lifecycle messaging and Segment for event tracking, so the agent was measured like any other feature of the product.
• Established and optimized the booking features: availability and scheduling, reservation flows and the client-facing experience on web and mobile, across the stack (Python, Ruby on Rails, React).

BairesDev

Senior Software Engineer · Remote

Engineer assigned to two products with nothing in common in the same year: live technical interviews and supply chain execution.

• 
  KalerisJul 2023 — Jan 2024

  Supply chain execution software: yards, port terminals and vessel operations. Integrations between systems that do not speak the same language.

• 
  CoderPadJan 2023 — Jul 2023

  Technical interview platform: two people write and run code at the same time, across dozens of languages. Product features and backend services.

• Two back-to-back assignments through BairesDev, each with its own domain and team: CoderPad and Kaleris.
• At CoderPad, the platform where technical interviews happen. Two people write and run code at the same time across dozens of languages, so an editor failure is not an ordinary bug: it takes down the interview that is happening. I worked on product features and the backend services behind them.
• At Kaleris, supply chain execution software for yards, port terminals and vessel operations. The work was in the integrations between systems that do not speak the same language, and in keeping cargo state traceable end to end.
• What repeats across both: unrelated domains, third-party integrations, and getting up to speed fast without breaking what already worked. The stack moved between Python, JavaScript and Ruby on Rails depending on what each team already ran.

Endava

Senior Software Engineer · Remote

Through Endava’s nearshore delivery, engineer assigned to Mastercard: internal APIs and backend components of its multitenant payment infrastructure, with Kafka, gRPC, ClickHouse analytics and an API designed to be consumed by a model.

• 
  MastercardOct 2021 — Jan 2023

  Multitenant payment infrastructure: internal APIs, financial event ingestion with Kafka and gRPC, an OAuth-secured API consumed only by the fraud model over a Neo4j graph, and ClickHouse analytics.

• Endava is an engineering consultancy with nearshore delivery out of LATAM and a payments practice going back more than twenty years. That is where the Mastercard assignment came from.
• Architected internal APIs and reusable backend components for the multitenant payment infrastructure. Built Python gateways and middleware layers for routing, rate limiting and tenant-scoped authentication.
• Contributed to a large-scale, low-latency system that ingested financial events in near real time using Kafka and gRPC. Developed Python consumers and producers with schema validation via Protobuf.
• For fraud and risk scoring we built an OAuth-secured API that only the model consumed: it got the OpenAPI spec in context plus a Neo4j graph with the base data, so it could interpret the relationships, reach its own conclusions and decide which endpoints to call. Credentials and endpoints came from environment variables per tenant and environment, so access was scoped and keys rotated without touching code.
• That was 2022, two years before MCP was published. It is the pattern MCP standardizes today, giving a model a described contract and letting it choose, and back then every company wrote its own version.
• Designed schemas for ClickHouse OLAP tables and built ingestion and transformation pipelines with Apache Airflow, plus React internal tooling to monitor transactions per tenant.

Globant

Software Engineer · Bogotá

Globant’s internal AI product, the line that later went to market as Augoor. Retrieval over code repositories and internal knowledge bases: indexing and embedding pipelines, hybrid search and answers anchored to their source. What people now call RAG, built in 2020.

• I worked on Globant’s internal AI product, on the two pieces the company took to its clients: Augmented Search and Augmented Coding. That line later became Augoor, the product Globant ended up taking to market.
• Augmented Search is retrieval over code repositories and internal knowledge bases. This was 2020 and 2021, before anyone called it RAG, but it is the same thing: I built the indexing and embedding pipelines, and the query APIs in Python.
• Ingestion came in through Ruby on Rails services that received the events, and processing ran distributed over Pub/Sub with Celery workers. Chunking followed code structure, on function and class boundaries, and every chunk carried its repository, path, commit and symbol. Reindexing was incremental against commits, not a full crawl every time.
• Retrieval combined keyword and semantic search, since pure vector search fails on exact identifiers, and filters were applied before the search rather than over the results.
• For grounding we returned the chunk with its origin and a link to the real source, and if nothing passed the threshold we returned nothing.
• Augmented Coding assists whoever is writing: autocompletion from the code that already exists, and documentation generated from the function itself. I also developed the React components for the A/B experiment dashboards, where model variants were compared before being switched on.

Mi Águila

Enterprise Solutions Architect · Bogotá

Platform modernization: monolith-to-microservices migration with tenant-aware service registration, Rails business logic (service objects, Sidekiq) and Kubernetes infra with Pulumi.

• Led platform modernization by migrating monolith to microservices, introducing tenant-aware service registration and auth boundaries. Rebuilt core business logic modules in Ruby on Rails with service-object patterns and background job processing via Sidekiq.
• Built React-based abstraction layers that enabled frontend/backend teams to extend business features across regions, including configurable UI components for multi-region fleet management dashboards.
• Managed Pulumi-based Kubernetes infrastructure across staging and production clusters.

LOGYCA

Lead Engineer · Bogotá

Internal platforms to unify API and identity: Python middleware (centralized logging, token validation), cloud migration with reusable Pulumi modules and a Rails admin.

• Focused on internal platforms to unify API and identity services for internal products. Developed Python-based middleware for centralized logging, authentication token validation, and cross-service request tracing.
• Drove cloud migration with a reusable Pulumi module system across development environments.
• Provided architectural input on internal authorization service and secrets management patterns. Built Ruby on Rails admin interfaces for managing service configurations and access control policies.

Universidad Piloto de Colombia

Software Developer · Bogotá

Maintenance and development of institutional systems with Ruby on Rails, PostgreSQL and Python; React interfaces for enrollment and academic scheduling.

• Supported and maintained institutional software systems. Delivered new development features under institutional standards using Ruby on Rails, PostgreSQL, and Python. Built internal React-based interfaces for student enrollment workflows and academic scheduling tools.