AI - Tech@Hoomanely (Page 3)

Retrieval as Infrastructure: Operating OpenSearch for AI Systems

For a long time, retrieval lived quietly in the background. It powered search boxes, filters, and internal tooling—important, but rarely the center of attention. AI changed that. In modern AI-backed products, retrieval is no longer an implementation detail. It sits directly on the critical path between user intent and

Boosting Edge AI with Latent Audio Generation

Edge AI is transforming how devices interact with the real world, from smart speakers to pet care devices. A critical challenge in deploying these models is robustness in varied acoustic environments. While collecting massive labeled datasets can help, real-world constraints often limit data availability, especially for specialised domains like pet

Augmenting Dog Eating Audio with Autoencoders

Introduction Audio is one of the hardest signals to work with in real-world ML systems. Unlike images, it is continuous, noisy, and deeply tied to the environment it is captured in. At Hoomanely, we build smart pet products that live inside homes - kitchens, living rooms, and feeding areas -

Building Unsupervised Anomaly Detection for Edge Audio

When your edge device captures thousands of audio clips in real-world conditions, the question isn't just "what's happening?" — it's "what's going wrong?" At Hoomanely, our Everbowl continuously monitors a pet to provide preventive healthcare insights for pets. But

Why Auto-Labeling Audio Needs Humans in the Loop

Machine learning promises automation at scale, but building reliable audio classification systems reveals a persistent challenge: ground truth validation. When we set out to build an edge-deployed dog sound classifier, we explored every path to automated labeling—and each one led back to the same conclusion: human perception isn'

How Do You Build Edge ML When You Don’t Have Labels Yet?

Introduction Building machine learning systems for the edge sounds exciting—until you realize that most real-world edge problems begin with a simple, uncomfortable truth: you don’t have labeled data. Not enough of it. Not the right kind. And certainly not data that reflects the messy environments your devices will

Beyond Traditional Augmentation: Training Edge Models for Real-World Chaos

When your AI model performs flawlessly in testing but stumbles in production, you're likely facing one of computer vision's most persistent challenges: domain shift. At Hoomanely, we encountered this exact problem with our dog segmentation model deployed on the Everbowl, our edge device designed to monitor

Learning Meals with Image Similarity

Intro One of the hardest problems in smart pet devices isn’t sensing - it’s understanding context. At Hoomanely, our EverBowl feeding system can reliably detect when a dog is eating using weight sensors, proximity sensing, and on‑device vision. But knowing what the dog is eating is a

How We Reduced False Detections by Training on “Bad” Indoor Data

Introduction Landmark detection models often look impressive during offline evaluation - clean datasets, sharp images, balanced lighting, and tidy annotations. But the real world is rarely that polite. At Hoomanely, we deploy dog‑face landmark detection models (eyes, nose, facial keypoints) on always‑on indoor devices mounted on feeding bowls.

Edge vs Cloud Inference: Finding the Right Balance for Real-World ML

Introduction Over the past decade, cloud inference has powered most large-scale machine learning systems — from recommendation engines to speech assistants. But as devices at the edge get faster, cheaper, and more privacy-aware, a hybrid pattern is emerging: compute just enough on the device to protect the user and reduce noise,