AI Engineer

About Stochastic

Most AI systems are built as one-size-fits-all solutions. Powerful in benchmarks, brittle in production. In a real workplace, they fall short. Not because they lack intelligence, but because they don't know your business. They're missing your domain knowledge, your workflows, your team's preferences, and the unwritten rules that make your organization run. That gap between general intelligence and operational context is where most enterprise AI implementations fail.

Stochastic was built to close that gap. We give enterprises a private, tailored AI system they fully own. One that learns continuously from real behavior and user feedback, adapts to their specific environment, and gets sharper the longer it's in use. Because the system is theirs, so is the data. No shared model training, no compromised proprietary information. Just an AI that knows your organization as well as your best employee does, and keeps getting better.

Our systems are already live in production, with a current focus on healthcare providers, one of the most regulated and context-dependent industries there is. Our agents handle inbound calls, fax processing, prior authorization, and the administrative work that pulls clinical staff away from patient care. It's a proving ground for the kind of high-stakes, high-context AI we're building toward at scale. Stochastic started with a group of Harvard AI systems researchers who built the first Bayesian and LLM inference accelerators, along with a real-time speech and NLP engine fast enough to hold a live conversation. The team has since grown to include engineers and researchers from Stanford, CMU, UIUC, NVIDIA, and Meta, all building toward the same goal: AI that belongs to your organization, not to anyone else.

About the Role

As an AI Engineer, you'll build Stochastic's multi-modal AI agents and train specialized models using reinforcement learning to make them excel at real healthcare workflows. This is a high-impact role shaping the core intelligence behind our product, including ownership of agent pipeline design, model fine-tuning, and acceleration.

Responsibilities

• Lead the design and development of agent pipelines, ensuring efficiency, scalability, and reliability
• Build and optimize multi-modal AI agents and ML pipelines, with emphasis on accelerating deep learning models
• Fine-tune, accelerate, and deploy LLMs within our production pipelines
• Develop and run RL training pipelines to specialize models for specific healthcare workflows
• Conduct research and experiments on the latest techniques for fine-tuning, acceleration, and agent pipeline optimization
• Build evaluation frameworks and benchmarks to measure agent performance
• Provide expert support for strategic customers on deployment and scalability challenges

Requirements

• Hands-on experience with Retrieval-Augmented Generation (RAG) systems and agent-based architectures
• Experience with reinforcement learning methods (RLHF, PPO, or similar)
• Strong proficiency in Python; experience fine-tuning models with PyTorch and the Transformers library
• Experience deploying deep learning models in production, optimized for efficiency and scalability
• Experience with at least one major public cloud provider (AWS, Azure, or GCP)
• Proficiency in Kubernetes for scalable deployment and orchestration of ML workloads

Strong Pluses

• 2+ years of experience in ML/AI engineering, with hands-on experience in model training and fine-tuning
• Experience accelerating models (quantization, distillation, latency optimization)
• Expertise in distributed systems and large-scale ML infrastructure
• Experience with Terraform for infrastructure as code
• Past experience as an ML engineer at a SaaS company
• Familiarity with Go for high-performance ML services