Get in touch ↗
PORTFOLIO · 2026SOFTWARE & MACHINE LEARNING ENGINEER

I build machine learning
systems for production.

Software engineering, distributed systems and high-performance ML inference.

From Python services and distributed processing to model serving, inference optimization and cloud deployment.

View engineering work ↓
01 · ABOUT

Where software engineering meets machine learning.

I'm Gonçalo Guedes, a Software & Machine Learning Engineer with 4+ years of experience building production ML systems.

My work goes beyond training models. I design the software, infrastructure and distributed systems required to run machine learning reliably at scale, from Python services and Kafka pipelines to GPU model serving with NVIDIA Triton and TensorRT.

I've architected systems processing 15M+ documents per month, improved production ML accuracy from 76.9% to 98.7%, deployed real-time models on constrained edge hardware and currently provide technical leadership to 9 Machine Learning Engineers. I also founded Chemolytic ↗.

04+Years building ML systems
15M+Documents / month
09ML Engineers led
E2EIdea to production
02 · EXPERIENCE

Systems I've built.

DEC 2025 — PRESENT
Glintt Global

Lead Machine Learning Engineer

AI/ML Chapter Lead

Hands-on technical lead for a chapter of 9 Machine Learning Engineers, taking GenAI, Computer Vision and NLP systems from R&D to reliable production services.

  • Lead architecture and implementation across ML systems, backend services and inference infrastructure.
  • Set engineering practices for code quality, model serving and inference performance.
  • Mentor engineers while remaining directly involved in technical design and delivery.
PythonPyTorchTritonTensorRTFastAPIDockerKubernetes
APR 2025 — DEC 2025
Glintt Global

Senior Machine Learning Engineer

Owned the architecture and deployment of a distributed Computer Vision and NLP system processing 15M+ documents per month.

  • Built Kafka-based pipelines that decoupled workloads for reliable, high-throughput processing.
  • Designed GPU model-serving infrastructure with NVIDIA Triton and TensorRT.
  • Developed Python and FastAPI inference services for downstream production systems.
  • Improved entity extraction accuracy from 76.9% to 98.7%.
15M+ / MONTH98.7% ACCURACY
PythonKafkaFastAPIPyTorchTritonTensorRTDockerKubernetes
JUN 2024 — APR 2025
Glintt Global

Machine Learning Engineer

Built and deployed a production RAG platform for enterprise information retrieval.

  • Developed Python backend services and APIs for ML-enabled applications.
  • Built document ingestion, indexing, retrieval and grounding workflows.
PythonOpenAILangChainWeaviateFastAPIRAG
FEB 2022 — JUN 2024
Cork Supply

Machine Learning Engineer

Developed the production ML system behind X100, an AI-powered X-ray tomography platform for non-invasive quality classification.

  • Replaced destructive laboratory testing cycles requiring up to 3 months with real-time inference.
  • Optimized deep-learning models for constrained edge hardware.
  • Achieved 6 inferences per second using TensorFlow Lite.
  • Took the system from applied R&D through optimization, deployment and product integration.
TensorFlowTFLiteComputer VisionEdge AIX-rayPython
03 · ENGINEERING

A model is only useful when the system around it works.

I approach machine learning as a software and systems engineering problem.

Data→Model→Service→Infrastructure→Production
MLSystems
01

ML Infrastructure & Performance

High-throughput model serving, inference optimization and production deployment.

GPU inference · batching · model serving · deployment
PyTorchTritonTensorRTONNXQuantization
02

Distributed Systems

Reliable asynchronous ML pipelines and backend infrastructure at production scale.

event-driven systems · decoupling · high-throughput processing
KafkaRedisQueuesPub/SubPython
03

Backend & Software Engineering

Production services and APIs that connect ML systems to real applications.

model APIs · web services · inference endpoints
FastAPIDjangoRESTPythonSQL
04

Cloud & Production Engineering

Containerized and scalable ML workloads across cloud and production environments.

Docker · Kubernetes · cloud deployment
DockerKubernetesAWSAzureCI/CD
04 · FOUNDER · PRODUCT ENGINEERINGEST. 2026 · PORTO

Chemolytic

chemolytic.com ↗

An ML product built from zero.

I designed and built Chemolytic end-to-end, combining scientific data workflows, backend engineering, machine learning and production deployment.

The platform turns raw spectroscopy data into production-ready predictive models through automated preprocessing, model selection, reproducible experiments, model management and deployable REST endpoints.

chemolytic.com
Chemolytic spectroscopy machine learning platform

Multi-format ingestion

FTIR, NIR, Raman and UV-Vis spectra from industry-standard formats and exports.

AutoML

Automated preprocessing, model selection and hyperparameter optimization across large experiment spaces.

Model registry

Versioned datasets and models with reproducible experiments and rollback.

REST API

Deploy trained models as secure inference endpoints.

Unsupervised explorer

PCA, t-SNE and clustering workflows for exploratory spectroscopy analysis.

Scientist mode

Full manual control over preprocessing, validation and hyperparameter optimization.

PythonDjangoPostgreSQLscikit-learnOptunaREST API
05 · RESEARCH

Engineering depth.
Applied research.

2026 · PUBLISHEDRead ↗

Calibration transfer between benchtop and handheld near-infrared instruments for predictive models of commercial milk powder samples using cubic spline interpolation, piecewise direct standardization and artificial neural networks

Chemometrics and Intelligent Laboratory Systems

NIR SpectroscopyCalibration TransferArtificial Neural Networks
2025 · PUBLISHEDRead ↗

Evaluating cork coating homogeneity using hyperspectral imaging and texture analysis

Journal of Near Infrared Spectroscopy

Deep LearningHyperspectralIndustrial AI
2023 · PUBLISHEDRead ↗

Handheld Near-Infrared Spectroscopy: State-of-the-Art Instrumentation and Applications

MDPI · Chemosensors

NIR SpectroscopyMachine Learning
06 · STUDY NOTES

Paper notes.
Built in public.

Deep learning papers unpacked through reading notes, visualisations and reproducible experiments.

Explore the repository ↗