Dublin, Ireland

AI Software Engineer

I build production AI systems, computer vision pipelines and robotics software. Currently at Moffett Automation. Previously safety-critical avionics at Airbus Defence and Space, UAV autonomy at Dronomy, and AI automation at Santillana (Sanoma Group).

MSc Robotics & Automation (UPM, thesis at TU Dortmund), one granted patent, three published papers.

Experience CV (PDF) GitHub

About

Most of my work is software: Python and C++ services, vision pipelines, and LLM-based systems that have to run unattended. At Moffett Automation I work on machine vision for industrial robotics and on AI-driven process automation.

Outside that I have designed and shipped three AI systems that are in production today — an agentic document inspection and orchestration service, a multimodal computer vision platform, and a workforce management system with vision-based check-in.

My research background is event-based vision. For my MSc thesis at TU Dortmund I built a self-supervised model that predicts 3D annotations for logistic objects from event-camera data. Before moving into AI I spent a year on verification and validation of DAL-A software for the Airbus MRTT, which is where I learned how much traceability and testing discipline matter.

Now: Robotics Software Engineer at Moffett Automation
Works with: Python · C++ · C# · FastAPI · Docker · Kubernetes · PyTorch · ROS2
Education: MSc Robotics & Automation (UPM) · BSc Electronics & Automation (UPM) · Advanced AI, Oxford (A+)
Output: 1 granted patent · 3 papers
Languages: Spanish (native) · English (C1) · German (A2)
Location: Dublin, Ireland · open to relocation

Results

Figures are approximate, measured against the process they replaced

~90%: Document review and update workload removed Moffett Automation
4mo → 2wk: Translation turnaround, at ~90% lower cost Santillana
98–99%: Potential saving versus historical translation budgets Santillana
~95%: Curriculum-adaptation cost cut, ~50% faster, brought in-house Santillana
+6.3 mAP: Object detection over the baseline MSc thesis · TU Dortmund
90%: Annotation storage reduction on the original dataset MSc thesis · TU Dortmund

Areas

Agentic AI & LLM systems

Multi-agent orchestration, tool and function calling, state and memory handling, human-in-the-loop review, guardrails and evaluation.

LangGraphLangChainCrewAI ReActMCP

Retrieval & LLMOps

Chunking and embedding strategy, vector stores, semantic retrieval, evaluation and observability, plus cost and latency tuning.

RAGVector Databases LLM EvaluationObservability

Computer vision

Detection, segmentation and 3D reconstruction, including event-based cameras, NeRF-style modelling and self-supervised training to reduce annotation cost.

PyTorchOpenCVYOLO ViTNeRF

Robotics & embedded

Perception and navigation for GNSS-denied UAV flight, collision avoidance and precision landing, and verification practice from certified avionics.

ROS2SLAM Sensor FusionDAL-A Embedded C/C++

Experience

Last 4 roles

Robotics Software Engineer

Feb 2026 — Present

Moffett Automation

Lead AI-driven process automation; cut document review and update workload by ~90% and reduced client-facing errors.
Build and integrate machine vision features and pipelines for core robotics processes.

AI & Automation Consultant

May 2024 — Jul 2026

Santillana (Sanoma Group)

Cut translation turnaround from ~4 months to ~2 weeks and costs by ~90%.
Built AI-ready datasets from editorial content using RAG, chunking, LLM-based tagging and knowledge graphs.

Robotics Engineer

Oct 2025 — Jan 2026

Dronomy

Perception and autonomy algorithms for UAVs in GNSS-denied environments.
Collision avoidance and precision landing for autonomous navigation.

Critical Embedded Software Engineer

Feb 2023 — Feb 2024

Capgemini Engineering · Airbus Defence and Space

Verification and validation of DAL-A safety-critical production software for the Airbus MRTT.

Full detail and production systems

Systems in production

Agentic document inspector

In production

Sales & engineering documentation

Flags contract-critical requirements and specification changes across large document sets, then propagates updates to every dependent document.

Real estate automation

In production

Multimodal computer vision

Generates portal-ready listings from structured data, with room classification, alternate-view generation, virtual staging and pre-valuation.

Workforce management

In production

Android & iOS · kiosk check-in

Time tracking with facial-detection check-in, shift and rest-interval monitoring, report automation and a digital fire register.

Research

Patent P202331053

Granted

Communicated system for detecting impacts and the status of an individual. Granted to Universidad Politécnica de Madrid, from my Bachelor's thesis on wearable capacitive sensing.

Patent and papers

Event-based 3D prediction

MSc thesis at TU Dortmund (FLW Chair): self-supervised prediction of multiple logistic objects with 3D annotations from event cameras. +6.3 mAP over baseline, 90% less annotation storage.

Stack

Full inventory on the CV page

Languages & tooling

PythonC++C# Embedded C/C++FastAPIREST APIs PydanticGitDocker KubernetesCI/CDMicroservices

Agentic AI & LLM

LangChainLangGraphCrewAI Multi-Agent SystemsAgent Orchestration Tool / Function CallingState & Memory ReActHuman-in-the-LoopMCP OpenAIAnthropicDeepSeek Open-Source LLMs

Retrieval & LLMOps

RAGAgentic RAGEmbeddings ChunkingVector DatabasesSemantic Retrieval Knowledge GraphsContext Engineering LLM EvaluationObservabilityGuardrails Prompt EngineeringCost & Latency Optimization

ML & computer vision

PyTorchOpenCVYOLO Deep LearningTransformersVision Transformers Diffusion ModelsGANsVAEs CLIPNeRF Self-Supervised LearningObject Detection SegmentationEvent-Based Vision

Robotics & safety-critical

ROS2GazeboAerostack2 SLAMLiDARSensor Fusion Path PlanningKalman & EKF Obstacle AvoidanceUAV Autonomy DAL-AVerification & Certification RTOSFPGA

Contact

Email is the most reliable way to reach me. I am employed and not actively looking, but open to talking about applied AI, agentic systems, robotics and industrial research.

tsvillaluenga@gmail.com LinkedIn More ways to reach me