Open to opportunities

Michael Nganga

@michaelnganga

Message

Pragmatic AI specialist delivering production-ready deep learning solutions.

Kenya

Message

What I'm looking for

I seek hands-on roles building production ML systems, with collaborative teams, clear impact metrics, and opportunities to lead model-to-prod delivery.

I am a Pragmatic AI specialist with 6+ years designing and deploying deep learning solutions across computer vision and natural language processing, with a strong record of moving research into production and delivering measurable business impact in Japanese and global markets.

I've led production deployments that reduced false negatives and latency, implemented automated model monitoring and retraining workflows, and mentored engineering teams while collaborating with product and sales to convert pilots to enterprise contracts.

Experience

Work history, roles, and key accomplishments

Current

Senior AI Specialist

Current

Preferred Networks

Apr 2022 - Present (3 years 10 months)

Led development and production deployment of an industrial anomaly-detection system reducing false negatives by 48% and saving clients ~¥120M annually; architected a multimodal PyTorch/ONNX pipeline that cut edge inference latency by 35% and implemented automated monitoring/retraining workflows reducing model drift incidents by 60%.

PyTorch ONNX Runtime Contrastive Learning MLOps Kubernetes Kubeflow Model Monitoring

Machine Learning Engineer

Google Japan

Jun 2019 - Mar 2022 (2 years 9 months)

Adapted transformer architectures for Japanese NLP to improve intent-classification accuracy by 22% and optimized multilingual pretraining to reduce compute costs by 28%; deployed BERT-derived models to Cloud AI Platform with CI/CD, cutting time-to-production by 40%.

Transformers NLP BERT CI CD

Research Scientist

Sony AI

Apr 2017 - May 2019 (2 years 1 month)

Developed real-time object tracking with an attention-based tracker improving mean IoU by 12% and collaborated on integrating models into power- and compute-constrained prototypes; authored dataset labeling and augmentation best practices reducing annotation cost by 30%.

Computer Vision Attention Mechanisms Data Labeling

Education

Degrees, certifications, and relevant coursework

The University of Tokyo

Master of Science, Computer Science (Machine Learning)

2015 - 2017

Completed a Master of Science in Computer Science with a focus on machine learning and a thesis on representation learning for sequence-to-sequence models applied to Japanese NLP.

Kyoto University

Bachelor of Science, Information Science

2011 - 2015

Completed a Bachelor of Science in Information Science with coursework in algorithms, statistics, and signal processing and an undergraduate project on feature extraction for visual recognition.