Role overview
Our client is an early-stage, venture-backed technology startup building the next-generation operating system for commercial HVAC suppliers and contractors. In just five months, they launched V1 of their first product and grew revenue 10x. The company is backed by top-tier investors and leading angels in the HVAC and industrial tech sectors, and is now focused on perfecting that product and scaling rapidly through 2025.
They’re looking for a
Deep Learning Engineer
with extensive experience in modern neural network techniques and PyTorch to help push the boundaries of computer vision in real-world environments. You’ll be joining a small, highly capable team focused on delivering practical, production-ready ML systems—from data pipelines through to fine-tuned models—in a fast-moving startup environment.
This role is ideal for someone who enjoys working with models under the hood, building and adapting training workflows, and applying research ideas to novel engineering challenges. The work involves more than model inference—you’ll design training workflows, develop evaluation pipelines, and engineer solutions that go beyond standard model usage.
What you'll work on
- Model Training & Development: Design and train deep learning models for layout analysis, OCR, object detection, image-to-graph, and related tasks. In some cases, you’ll extend or adapt existing architectures; in others, you’ll help design custom approaches from the ground up.
- Evaluation and Monitoring: Build robust metrics, monitor production model performance, and proactively identify failure modes and areas for improvement.
- Collaboration and Integration: Work closely with the engineering team to integrate models into the company’s product and infrastructure. Participate in architecture and roadmap decisions.
What we're looking for
- Experience with active learning setups
- Applied experience with RLHF (Reinforcement Learning from Human Feedback)
- Published research developing SOTA computer vision or other DL models
- Experience with deployment and monitoring pipelines for ML systems