DecenTrialz is a clinical trial discovery and recruitment platform. We are building recruitment and engagement SaaS, a trial marketplace and personalization AI that help patients find the right trials and help sponsors and sites recruit faster. We move quickly and ship real AI features to production.
What you will do
Design, build and ship AI agents and agentic workflows that solve real product problems
Prototype fast, validate quickly and take working solutions to production
Build multi-step agent workflows with tool calling, state management and orchestration
Build RAG pipelines: document processing, chunking, embeddings and retrieval over trial and medical data
Develop backend APIs and services in Python using FastAPI
Integrate and switch between LLM providers (Azure OpenAI, OpenAI, Anthropic and others) based on cost and quality
Add evaluation, guardrails, structured outputs and monitoring to reduce hallucinations and control inference costs
Own features end to end, from design to deployment and production support
What we are looking for
Strong Python skills and experience building production backend services
Hands-on experience building AI agents using LangChain, LangGraph or similar frameworks
Solid understanding of LLMs: prompting, context windows, embeddings, tool calling and generation parameters
Experience building RAG systems and working with vector databases
Experience with FastAPI, async programming and API integrations
Working knowledge of Azure or AWS, including deploying containerized applications
Comfortable with relational and NoSQL databases (PostgreSQL, MongoDB)
A bias for speed: you ship working solutions fast and improve them in iterations
Ability to work independently across backend, AI and cloud layers
Positive to have
Experience with Azure AI Foundry and Azure OpenAI
Experience with Elasticsearch or OpenSearch
Familiarity with CI/CD, Docker, Terraform or Bicep
Data processing experience with Pandas or Polars
Exposure to healthcare data, clinical trials or compliance standards like HIPAA
Experience with fine-tuning, PEFT or LLM observability tools