We are not looking for someone who knows how to call OpenAI's API and call it AI.
We are looking for an engineer who has rolled up their sleeves — trained models from scratch or fine-tuned foundation models on custom datasets, debugged loss curves, written custom training loops, and shipped something that runs in production.
What you will work on
At Perimattic, you will be building AI systems that go inside real enterprise products — document intelligence, manufacturing analytics, and agentic workflows. You will work on model selection, fine-tuning, evaluation pipelines, and inference optimization. You will own the full lifecycle from problem framing to deployment.
What we expect from you
- 3 to 4 years of hands-on experience in ML engineering, not AI product management
- You have trained or fine-tuned models — transformers, CNNs, or similar — on real datasets with real constraints
- Solid command of PyTorch or TensorFlow, not just the scikit-learn docs
- You understand what overfitting actually looks like in a training run, not just in theory
- Experience with model evaluation, dataset curation, and iterative experimentation
- Comfort working with unstructured data - documents, images, or time-series
- Bonus if you have worked with LLM fine-tuning (LoRA, QLoRA, PEFT), RAG architectures, or ONNX/TensorRT for inference
- Bonus if you have deployed models to production not just notebooks
What you will not find here
- A job where your entire AI work is chaining LangChain calls
- A company that confuses API integration with machine learning
- A team that won't let you get your hands dirty
What you will find
- A 13-year-old product and services company that is transitioning aggressively into AI-first products
- Real problems across manufacturing, document processing, and enterprise automation
- A team that values depth over buzzwords
- Room to grow into a senior or lead role as we scale our product portfolio
