Machine Learning Engineer

Staffpulse Technologies is a YC startup focused on AI-powered CAD solutions, and they are seeking a Machine Learning Engineer to oversee the lifecycle of machine learning systems. The role involves deep research and engineering tasks, collaborating with founders, and designing custom deep learning models to enhance CAD workflows.

Responsibilities:

• Design, train, and iterate on custom deep learning models that understand CAD workflows and predict high-quality next-step suggestions
• Build and maintain robust Python training and evaluation pipelines, including data preprocessing, experimentation, and offline/online metrics
• Architect model serving and backend components so that features are fast, reliable, and easy to integrate into CAD environments
• Work closely with the founder and early customers (mechanical / hardware engineers) to understand real-world workflows and translate them into ML formulations
• Own the full lifecycle of ML features—from research and prototyping through productionization, deployment, and monitoring
• Collaborate with the broader engineering team on core product and infrastructure work (backend, APIs, data models, performance)
• Establish best practices for experimentation, logging, and model comparison to ensure steady improvements over time
• Stay current on relevant ML research (e.g., sequence models, geometric deep learning, representation learning) and decide pragmatically what is worth applying

Requirements:

• Deep ML expertise – 4+ years of hands-on machine learning experience (or equivalent research/thesis-based Master's or PhD), with a track record of training and improving deep models, not just using pre-built APIs
• Strong Python engineering – you write clean, well-structured, production-ready Python without handholding, including tests, documentation, and thoughtful abstractions
• End-to-end ownership—experience owning ML systems from data to deployment: building training pipelines, running experiments at scale, tuning hyperparameters, and shipping models into real products
• Applied problem-solving – proven ability to take messy, open-ended product requirements and turn them into concrete ML formulations, experiments, and shipped features
• Collaboration & communication – able to work closely with founders, engineers, and (eventually) customers; can explain trade-offs and model behavior clearly to both technical and non-technical partners
• Startup mindset – comfortable in a fast-moving, low-process environment; willing to wear multiple hats across research, engineering, and backend work when needed
• Demonstrated experience designing custom architectures, writing training loops, and shipping models you built from scratch (not fine-tuning or prompting existing models)
• Strong proficiency in Python and at least one deep-learning framework (PyTorch preferred)
• 4+ years of industry or equivalent academic experience working on machine-learning systems
• Based in the US with existing work authorization, Hestus requires a US citizen/visa only for this role
• Published research or meaningful open-source contributions demonstrating novel technical work?
• Expert with PyTorch (preferred) or similar frameworks such as TensorFlow / JAX; comfortable implementing and modifying custom architectures, loss functions, and training loops
• Experience with geometry / graphics / CAD, 3D representations, or robotics; familiarity with cloud ML platforms (AWS / GCP) and backend frameworks (Flask, FastAPI, Django)
• If available, please share a link to your GitHub, portfolio, or any recent projects you've worked on