ML Engineering • Data Systems • MLOps
Yohan Shanuka
Engineering scalable data pipelines, MLOps workflows, and cloud-native ML systems.
Engineering Mindset
Building Scalable ML & Data Systems.
I focus on building intelligent systems at the intersection of machine learning and data engineering, designing high-throughput distributed pipelines and deploying production-ready models that solve complex real-world challenges.
My goal is to develop production-ready machine learning workflows supported by reliable data infrastructure, modern backend systems, and scalable cloud architectures.
Engineering Focus
Core Expertise
Three focused areas — each with a clear pipeline and the capabilities I bring to production systems.
ML Engineering
Train, evaluate, and serve models with low-latency APIs.
- CNN & transfer learning
- Model APIs & FastAPI
- Prediction systems
- Model optimization
Key Performance Index
Inference
Tools & Frameworks
Technology Ecosystem
A curated stack I use to build scalable ML systems, data pipelines, and cloud-native infrastructure.
Machine Learning
Ecosystem FocusDeveloping and deploying deep learning, computer vision, and predictive models using modern frameworks.
Core Competencies
Ring % = self-assessed proficiency · actively expanding the stack
Case Studies
Production AI Systems
Deep-dive into the architecture, challenges, and metrics behind intelligent systems I've engineered.
End-to-end streaming pipeline handling high-throughput event data with sub-500ms latency. Replaced legacy batch jobs that caused 24-hour reporting delays.
24-hour delays in critical metric reporting due to batch processing.
End-to-end streaming pipeline handling high-throughput event data with sub-500ms latency. Replaced legacy batch jobs that caused 24-hour reporting delays.
24-hour delays in critical metric reporting due to batch processing.
IoT-driven pipeline combining streaming sensor ingestion with ML prediction to monitor livestock health patterns and trigger real-time alerts.
Manual monitoring caused late disease detection and yield loss.
IoT-driven pipeline combining streaming sensor ingestion with ML prediction to monitor livestock health patterns and trigger real-time alerts.
Manual monitoring caused late disease detection and yield loss.
Production-grade CNN microservice for automated crop disease diagnosis. Sub-100ms inference latency with a fully containerized deployment pipeline.
Farmers needed reliable automated API for rapid field image diagnosis.
Production-grade CNN microservice for automated crop disease diagnosis. Sub-100ms inference latency with a fully containerized deployment pipeline.
Farmers needed reliable automated API for rapid field image diagnosis.
Research & Publications
Peer-reviewed research at the intersection of IoT systems, edge AI, and real-world agricultural engineering challenges.
Real-Time Cattle Monitoring Using Low-Cost IoT Smart Collars with LoRa Communication in Sri Lanka's Dry Zones
Authors
Abstract
Explores the design and deployment of a low-cost IoT smart collar system using LoRa communication for real-time cattle health monitoring across remote dry-zone environments in Sri Lanka. The system captures biometric and behavioural data — including temperature, motion, and heart rate — and transmits it over long-range low-power networks to an ML-backed prediction and alert pipeline.
More research in progress — stay tuned for upcoming publications.
System Design
Scalable Architecture
Blueprint-level architecture diagrams showcasing distributed workflows, ETL patterns, and containerized deployments.
Streaming Data Architecture
High-throughput event pipeline
Select any component in the blueprint diagram on the left to initialize logs and specs telemetry.
Active Development
Currently Building
Real-time progress on active engineering projects — because great systems are always evolving.
Cattle Health AI Monitoring
Real-time IoT sensor ingestion + CNN-based anomaly detection with alerting pipeline.
CNN Tomato Disease Classifier
Transfer learning pipeline with automated retraining triggers and model versioning.
Streaming ML Inference Pipeline
Sub-100ms inference pipeline serving Kafka-triggered predictions via FastAPI endpoints.
Commit Activity
Last 7 months
Certifications
Professional credentials validating expertise in cloud architecture, machine learning, and data engineering.
AWS Certified Machine Learning – Specialty
Data Engineering with Google Cloud
Deep Learning Specialization
How I Think
Engineering Philosophy
I focus on building scalable, production-ready AI systems that combine machine learning, distributed data pipelines, and efficient backend infrastructure — not just models, but intelligent systems.
Scalability
Systems that grow with demand. Every pipeline I build is designed to handle 10× the expected load from day one.
Reliability
99.9% uptime is the baseline. I engineer fault-tolerant architectures with graceful degradation and self-healing capabilities.
Observability
You can't improve what you can't measure. Every system ships with metrics, logs, and alerting baked in from the start.
Get In Touch
Looking for an MLOps or Data Engineer to build scalable systems? Let's connect.