Backend AI Engineer

Spectraforce

United States

6 days ago

Apply on leoforce.us

Prepare application

Role details

Contract type

Temporary contract

Employment type

Full-time (> 32 hours)

Experience level

Experienced

Experience required

4 years minimum

Working hours

Regular working hours

Job source

leoforce.us

Tech stack

Application Programming Interfaces (APIs) Artificial Intelligence Amazon Web Services Data Analysis Data Infrastructure Data Mining Software Design Documents Design of User Interfaces Python (Programming Language) PostgreSQL Machine Learning Performance Tuning

+19 more

Next.js Unstructured Data WebSocket Pytorch ReactJS Flask (Web Framework) Large Language Models Prompt Engineering Backend Fastapi Containerization AI Platforms Kubernetes Machine Learning Operations Front End Software Development Restful APIs Streamlit Framework Docker Microservices

Job description

RAG & Prompt Engineering: Craft and refine effective prompts for RAG, grounding, and context tuning to achieve optimal AI performance in product development.
High-Performance API Engineering: Develop asynchronous microservices (FastAPI) using Server-Sent Events (SSE) or WebSockets to stream real-time LLM responses to front-end UIs without backend timeouts.
Vector Database Infrastructure: Design, develop, and implement robust Vector Databases using LLMs and modern retrieval technologies to capture information from diverse engineering sources (PDFs, design docs, regulatory guidelines).
Data Extraction & Structuring Pipelines: Build and optimize pipelines to extract and structure multi-modal data (tables, text, images) from unstructured documents for LLM training, grounding, and runtime query execution.
LLM Fine-Tuning & Training: Fine-tune and train generative AI models using Dexcom’s engineering data and domain knowledge to create high-accuracy, domain-specific models.
GenAI Application & Tool Development: Design and implement scalable backend APIs (FastAPI/REST) and UI integration interfaces so internal engineers can query knowledge bases and analyze data.
Automated Requirements Generation: Develop backend functionalities to automatically generate technical requirements from design documents, user stories, and system specification files.
Documentation & Knowledge Transfer: Thoroughly document architecture, code, REST endpoints, and model training procedures to enable seamless knowledge transfer to Dexcom internal teams.
Cross-Functional Collaboration: Partner closely with Subject Matter Experts (SMEs), System Engineers, and V&V Test teams to optimize AI-powered workflows.
AI Guardrails, MLOps & Cost Governance: Implement hallucination checks, PII masking, and guardrails (e.g., NeMo Guardrails) for medical device context. Track token usage, latency, and costs using LangSmith or Vertex AI monitoring.

Requirements

Relevant Experience & STEM Foundation: 4+ years of professional software/ML engineering experience, with a dedicated AI/ML focus in the last 1-2 years.
Google Gemini / Vertex AI (Non-negotiable): Hands-on experience with the Gemini model family and Vertex AI, including deployment, grounding, and integration into production AI services. Hands-on experience with containerization (Docker) and deploying services via Cloud Run or GKE (Kubernetes).
Languages & AI Libraries: Proficiency in Python and modern ML/AI frameworks (PyTorch, LangChain, LangSmith) for building autonomous LLM agents, tools, and RAG pipelines.
Agent Building & Tool Calling: Proven experience building AI/LLM agents and tool-calling systems in Python against unstructured, multi-source data.
Context Engineering & RAG: Expertise in RAG pipelines, prompt engineering, context tuning, grounding, and Vector Databases (e.g., Milvus, Postgres/Pgvector). Clear understanding of advanced RAG architecture including Hybrid Search (Vector + Keyword), Re-ranking models, and semantic caching.
Unstructured Data Handling (Non-negotiable): Demonstrated ability to ingest, clean, extract, and structure text, tables, and images from unstructured documents (PDFs, design docs, regulatory files) for LLM training and usage., * Direct experience designing and deploying high-throughput REST APIs (e.g., FastAPI/Flask).
Familiarity with medical device development regulations and compliance (e.g., FDA guidelines, ISO 13485).
Experience integrating multiple LLM APIs beyond a single vendor (e.g., OpenAI, AWS Bedrock, Anthropic Claude).
Front-end development/integration experience for UI design (e.g., Streamlit, Gradio, React/Next.js integration).