Sergio Jimenez Jimenez
PROFESSIONAL SUMMARY
Senior engineer with around ten years building production systems, much of it for regulated industries. At Pdftools I own the AI enablement platform across product lines and have shipped on Smart Redact (AI-based PII removal for healthcare, finance, government, and legal) and OCR Service (horizontally scalable manager / worker architecture for 180+ languages). On the side, Liftory is my personal R&D platform: an agentic chat worker on the same scaling patterns, used to validate ideas before they reach the day job. Track record of setting SDLC standards, mentoring engineers, and translating research into reliable services.
WORK EXPERIENCE
- Own the AI enablement platform consumed across Pdftools product lines: ONNX Runtime model serving, FastAPI inference services, and NLP pipelines on Hugging Face Transformers and spaCy.
- Take models from research to production end-to-end (PyTorch / TensorFlow training, ONNX export, containerized Kubernetes inference) with CI/CD, observability, and regression evaluation built into the release pipeline.
- Partner with product and research engineers to translate model research into reliable services consumed by multiple product teams.
- Core engineer (team of 4 within a 20+ engineer Swiss organization) on Smart Redact, the AI-based PII removal product for regulated industries (healthcare, finance, insurance, government, legal). Hybrid Named Entity Recognition plus deterministic rules across seven languages.
- Also contributed to OCR Service (180+ languages). Both products run on AWS on a horizontally scalable manager / worker architecture I helped design.
- Own infrastructure as code (Terraform), CI/CD (Azure DevOps and GitHub Actions), observability, and on-call practices across product lines.
- Drove the team's SDLC standards (review norms, testing strategy, release automation, architecture decision records) and mentor engineers across teams.
- Architected the Grand Solver: a real-time cost-evaluation system used by operations crews across multiple airlines in the Lufthansa Group to make live operational decisions. Cloud-native on GCP, delivered in partnership with Google Cloud.
- Promoted to lead on the Lufthansa Group-wide OPSD initiative. Built simulation and automation components for cross-airline cost comparison, contributed to the long-term roadmap, and ran cross-airline refinement sessions.
- Recognized internally as an outstanding high achiever for technical depth, stakeholder communication with non-engineering domain experts, and collaborative mindset.
- Full-stack cost-maintenance and budgeting application reconciling vendor invoices against contract terms. Stack: .NET Framework and .NET Core backend, Angular with RxJS frontend, PL/SQL on Oracle, Docker on GCP.
- Senior C++ developer on an exchange platform running on Linux: CMake, Docker, Redis, RabbitMQ, Google Protocol Buffers, gtest and gmock. Profiled with Valgrind and perf. Services containerized on Amazon EKS, with CI on Jenkins and TeamCity.
- ERP for tourism clients (including SkiData): JEE8 on WildFly with an Ext JS frontend. Tooling: Maven, Arquillian, Jenkins.
- Warehouse management system modules built on .NET Web API, WPF, WinForms, and Entity Framework. Designed and built a 3D warehouse-simulation tool (XNA / MonoGame embedded in WPF) that lets users configure a layout, persist it to the database, and run virtual operations end-to-end.
- BDD with SpecFlow, unit testing with NUnit and Moq, and a CruiseControl.NET, Nant, and WiX release pipeline.
- Android and Android Wear application in Java for Sweetspot. The application won the company-hosted contest. Earlier, computer-systems maintenance at the University of Oviedo.
PROJECTS
- Founder and sole engineer. Shipped end-to-end: an Ionic and Angular mobile app on the App Store and Google Play, native watchOS (Swift) and Wear OS (Kotlin and Jetpack Compose) companions with bidirectional phone-to-watch sync, a Next.js marketing site at liftory.ch, and a .NET 10 backend on PostgreSQL.
- Built an agentic AI chat worker (skill-driven, multi-turn) that designs personalized training plans. RAG over a curated training and nutrition corpus using a local ONNX bge-m3 embedding model and hybrid BM25 plus cosine retrieval, orchestrated via OpenAI's gpt-5-mini with constrained decoding for typed tool calls. Chose RAG over fine-tuning: the corpus updates frequently and is too small for fine-tuning to add value.
- Workers are stateless and horizontally scalable behind RabbitMQ, mirroring the manager / worker pattern I work on at Pdftools. Full OpenTelemetry observability (Grafana, Loki, Tempo, Prometheus) and a private Gitea content pipeline publishing versioned brand-asset packages consumed by the web and mobile builds.
EDUCATION
SKILLS
CERTIFICATIONS
- Azure Cloud Developer (https://www.credly.com/badges/03a223d7-c425-4fc9-b7b4-2a0e4aa57471)
- SAFe 4 Agilist (https://www.youracclaim.com/badges/939031f8-f301-4e6c-9bf3-70033df29c3e/public_url)
- Scrum Master (https://www.scrum.org/certificates/495301)
LANGUAGES
- English: C1
- Spanish: Native
- German: A2 / B1