Sr AI Engineer

Your Mission

To architect a secure, hybrid AI ecosystem that combines the power of Large Language Models (LLMs) with the privacy of local inference. You will build the "brain" of the platform: a sophisticated RAG (Retrieval-Augmented Generation) engine that indexes millions of words, understands user intent, and delivers personalized answers without ever exposing sensitive data.

Role Overview

We are seeking a pragmatic and security-minded AI Engineer to lead the development of our Generative AI and Personalization modules. This is not a role for someone who simply writes prompts for ChatGPT.

You will be engineering a Hybrid AI Architecture: deploying open-source models (e.g., Llama, Mistral) for private, local data processing, while securely orchestrating external LLM APIs (OpenAI/Anthropic) for public-facing tasks. You will own the Vector Database, the Semantic Search algorithms, and the critical Data Safety Layer that prevents PII leakage. You will leverage an AI-first workflow (using tools like Cursor and Claude) to rapidly prototype and deploy these complex systems.

Key Responsibilities

• Engineer RAG Pipelines: Architect and build the Retrieval-Augmented Generation system that powers the site's search. This involves chunking, embedding, and retrieving context from a dataset of ~38 million words.
• Build Private Inference Models: specific content classification tasks (e.g., auto-tagging legacy documents), deploy and optimize open-source Small Language Models (SLMs) and LLM to run entirely within our private cloud infrastructure to ensure data privacy.
• Vector Database Management: Manage the vector search infrastructure (e.g., Pinecone, Weaviate, or Milvus), optimizing embedding models to ensure high relevance in search results.
• Implement AI Safety & PII Scrubbing: Build the "airlock" systems that automatically detect and redact Personally Identifiable Information (PII) from user queries before they are sent to any third-party model .
• Vector Database Management: You are responsible for managing the vector search infrastructure (e.g., Pinecone, Weaviate, or Milvus) and optimizing embedding models to ensure high relevance.
• Develop Personalization Algorithms: Engineer privacy-safe recommendation engines that analyze anonymized behavioral data (collaborative filtering) to surface relevant content based on user roles .
• MLOps & Model Governance: Implement monitoring to detect "Model Drift" and automated pipelines for re-indexing content when new documents are published .
• AI-Accelerated Development: Actively utilize AI coding assistants (Cursor, Claude CLI, Gemini) to write boilerplate Python code, generate complex SQL/Vector queries, and unit test data pipelines.

Core Skills and Competencies

• Tech Stack: Expert-level Python skills. Proficiency with AI frameworks like PyTorch or TensorFlow.
• Generative AI Frameworks: Deep experience with orchestration libraries like LangChain or LlamaIndex.
• Vector Search: Hands-on experience with Vector Databases and embedding models (e.g., OpenAI text-embedding-3, HuggingFace models).
• Hybrid Architecture: Ability to work with both 3rd Party APIs (OpenAI, Anthropic) AND Local/Open Source Models (Ollama, vLLM, HuggingFace Transformers).
• Data Engineering: Experience building ETL pipelines to ingest unstructured data (PDFs, HTML) for AI processing .
• Security: Understanding of data redaction techniques and secure API handling.

Required Tech Stack

This role requires a "Hybrid Polyglot" who can work with both commercial APIs and raw open-source model weights.

• Core Language: Expert-level Python.
• AI Frameworks: Deep experience with PyTorch or TensorFlow.
• Orchestration: Proficiency with LangChain, LangGraph or LlamaIndex.
• Vector Databases: Hands-on experience with Pinecone, Weaviate, or Milvus.
• Model Ecosystem (Hybrid):
• Public APIs: OpenAI, Anthropic.
• Local/Open Source: Ollama, vLLM, Hugging Face Transformers.
• Data Engineering: Experience building ETL pipelines for unstructured data (PDFs, HTML).
• AI Workflow Tools: Proficiency in Cursor and Claude CLI.
• Code versioning: Git

Domains of Mastery

• Retrieval Optimization: The ability to tune a search engine so it doesn't just find keywords, but understands concepts (Semantic Search).
• Privacy-Preserving ML: Knowing how to build personalization systems that track behaviors without tracking individuals (anonymization techniques) .
• Latency Optimization: Balancing the trade-off between the "smartness" of a model and the speed of the response (using caching and lightweight models where possible).
• Prompt Engineering (System Level): Writing robust system prompts that prevent hallucinations and enforce brand tone constraints.

Job Type: Full-time

Pay: $30, $45,000.00 per month

Work Location: Remote