About the Role

We are looking for an AI Engineer to join a small, fast-moving team building an AI-powered SaaS platform from the ground up. You will own the intelligence layer of the platform - the components responsible for language understanding, semantic search, embedding pipelines, and speech-to-text processing.

This is a full-time position.

What You Will Be Doing

• Building and maintaining a provider-agnostic AI abstraction layer supporting multiple LLM providers
• Designing and managing embedding generation pipelines with structured metadata filtering
• Integrating and maintaining a vector database for semantic search at scale
• Building and maintaining a speech-to-text transcription pipeline with quality validation and fallback handling
• Implementing semantic analysis logic to extract structured insights from conversational text
• Working from detailed written specifications provided by the product owner
• Submitting daily written progress updates

Must-Have Skills

• Python - 3+ years in production AI or ML systems
• OpenAI API - embeddings and chat completions in production (not just tutorials)
• Vector database experience - Weaviate, Qdrant, Pinecone, or equivalent in a production environment
• Embedding pipeline design - you understand embedding drift, dimensionality, and metadata filtering
• Speech-to-text integration - Whisper or equivalent in a production pipeline
• NLP or text analysis in deployed, production systems
• Building AI abstraction layers that are not hard-coded to a single provider

Nice to Have

• Experience with behavior or intent detection from conversational text
• Familiarity with LangChain, LlamaIndex, or similar orchestration frameworks
• Background in conversation intelligence, speech analytics, or call center technology
• Experience with Weaviate specifically including hybrid search
• Understanding of RAG (retrieval-augmented generation) architectures in production

Non-Negotiables

• Production systems only. Tutorial projects and demos do not qualify.
• You must design systems that can swap AI providers without rewriting core application logic. Hard-coding a single provider is a disqualifier.
• You must understand that embedding generation and vector storage are separate responsibilities.
• You must think about fallback handling, rate limits, and model output validation - not just the happy path.
• IP assignment agreement and NDA must be signed before engagement begins.
• You will not communicate directly with other developers. All coordination goes through the product owner.

Engagement Terms

• Competitive rate based on market levels for your region
• Contract with IP assignment and NDA

Job Types: Full-time, Permanent

Benefits:

• Health insurance

Application Question(s):

• A description of a production AI system you built - what it did, what the hard parts were, and what you would do differently today?
• Your GitHub profile or a code sample showing Python AI integration work?
• Your expected monthly rate and availability?

Work Location: Remote