> Markdown version of [/jobs/ext/4749457-ai-engineer-iii](https://www.wearedevelopers.com/jobs/ext/4749457-ai-engineer-iii). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # AI Engineer III - **Company:** Intone Networks - **Location:** United States (Remote available) - **Contract:** Permanent contract - **Skills:** Training Data, 3d Models, Artificial Intelligence, Data Files, Python (Programming Language), Machine Learning, Systems Development Life Cycle - **Published:** November 16, 2026 - **Apply:** https://www.careerbuilder.com/job-details/ai-engineer-iii--b5d181df-3d1d-47a8-8353-43d1380f5dc3 ## About the Role * Required * Strong experience in machine learning, computational chemistry, cheminformatics, or related scientific data science domains * Experience training and fine-tuning models on domain-specific scientific datasets * Experience with 3D modeling workflows, pose generation, or docking-based comparison studies * Working knowledge of benchmarking methodologies and model performance evaluation * Strong Python and scripting/automation skills * Ability to design reproducible, scalable model workflows with runtime efficiency in mind * Strong collaboration and communication skills across scientific and technical teams ## Description This is a 4-month contract role for an AI Engineer III, based remotely in the US. You will support the development, fine-tuning, and benchmarking of AI/ML models for computational drug discovery, building a scalable workflow that goes from protein and ligand sequence straight to 3D structure and affinity prediction with minimal manual intervention, across multiple lead optimization programs. Responsibilities * Fine-tune pre-trained foundation models (e.g. Boltz-2, AISB) on domain-specific project data * Compile datasets from multiple lead optimization projects, spanning experimental protein-ligand complex structures and potency/affinity data * Evaluate dataset variation to challenge model applicability domains * Benchmark model performance against established methods (DeepAutoQSAR, GatorAffinity, MPNN, MM-GBSA, FEP, etc.) * Compare co-folder pose generation vs. docking-based methods * Evaluate recall and precision using project-specific thresholds; build model ranking based on confidence and applicability domain * Determine whether one fine-tuned co-folder model can serve multiple projects or whether project-specific models are needed * Design an automated, end-to-end workflow optimized to generate single predictions in under a minute on GPU or CPU