Talkiatry is a national mental health practice focused on making high quality mental healthcare more accessible. They are seeking a Director of AI to lead AI innovations, define the AI technical roadmap, and collaborate with cross-functional teams to enhance the delivery of mental health services.
Responsibilities:
- Define and lead the AI roadmap across experimentation, model evaluation, deployment, and iteration
- Architect systems to support safe and scalable LLM interactions—including prompt chaining, memory retention, and RAG pipelines
- Develop strategies for model selection (e.g., Azure OpenAI, AWS Bedrock, open-source) and cost-effective inference at scale
- Design and implement experiments to evaluate the efficacy, safety, and impact of AI innovations
- Build and scale a high-performing AI team across machine learning, data science, MLOps, and backend infrastructure
- Serve as the AI technical authority in conversations with external vendors, cloud partners, compliance officers, and executive stakeholders
- Partner with legal and clinical teams to ensure HIPAA compliance, BAA alignment, and safe handling of sensitive patient data
Requirements:
- Expert in large language models (OpenAI, Anthropic, Mistral, etc.), including fine-tuning, prompt engineering, embeddings, and context window optimization
- Hands-on experience with RAG frameworks, vector databases (Pinecone, FAISS, Redis), and memory architectures
- Strong programming background in Python and ML stack (PyTorch, HuggingFace Transformers, LangChain, Weights & Biases)
- Proficiency with cloud platforms (AWS, Azure), containerization (Docker, Kubernetes), and orchestration tools (Airflow, FastAPI, etc.)
- Deep familiarity with model evaluation frameworks—automated and human-in-the-loop—for conversation safety, empathy, and hallucination detection
- 7+ years of experience in AI/ML roles, with at least 3 years leading cross-functional teams including engineering, product, and research
- A successful track record of shipping production-grade AI systems—ideally in healthcare, mental health, or another mission-critical domain
- Deep technical fluency in LLMs, retrieval systems, and safe interaction design—with an ability to lead architectural decisions and implementation
- Experience building or managing AI teams from 0→1, with a strong recruiting eye and mentorship skills
- Familiarity with ethical AI, including red-teaming, bias mitigation, explainability, and escalation protocols for risky outputs
- Strong understanding of healthcare compliance (e.g., HIPAA, PHI), including how to architect systems with auditability and patient safety in mind
- A pragmatic approach to experimentation and iteration—balancing cutting-edge AI with operational feasibility and clinical guardrails
- Exceptional communication skills—you can convey technical decisions to both engineers and executives, and lead strategic conversations with confidence