Vector Databases Jobs in Memphis, TN (NOW HIRING)

PURPOSE OF POSITION Responsible for model integration, data pipelines, retrieval infrastructure, and the engineering scaffolding required to ship reliable, secure, and cost-effective Artificial Intelligence (AI) features. This role ensures the delivery of production-grade Large Language Model (LLM) systems that meet real-world demands for performance, cost-efficiency, and governance. MINIMUM QUALIFICATIONS Education: Master's degree preferred.

Bachelor's in Computer Science, Data Science, AI, or related field with equivalent experience considered, or related field or equivalent practical experience. Training and Experience: 3-7 years in backend development, AI systems, or related roles, with a focus on LLMs integration or retrieval systems. General Skills: Must have strong software engineering fundamentals and a deep understanding of working with LLMs in production environments.

The ideal candidate brings hands-on experience with Python and modern data tooling and is comfortable building robust pipelines that connect unstructured content, structured data, and retrieval systems to power context-aware LLM workflows. You should demonstrate fluency in the design and reasoning of data movement processes, including ingestion, preprocessing, vector indexing, and query generation. Experience working with both open-weight and API-based large language models is also essential.

This role requires a practical mindset, a strong command of SQL and retrieval strategies over relational data, and the ability to experiment, evaluate, and iterate toward scalable, cost-effective, and trustworthy AI features. Required Skills: Mastery in Python, including experience with modern practices in structuring, testing, and maintaining codebases. Orchestrated Retrieval-Augmented Generation (RAG) systems, including document chunking, embedding, vector search, and grounded context construction.

Expertise with PostgreSQL and pgvector, including schema design and structured retrieval over relational data. Robust operational understanding with SQL query generation, particularly in the context of semantic or hybrid retrieval. Comprehensive background integrating and orchestrating LLMs, with a focus on prompt templating, tool usage, and response parsing.

Familiarity with Google ADK or equivalent frameworks for LLM scaffolding and orchestration. Proficient in utilizing unstructured and structured data, including ingestion from PDFs, DOCX, Markdown, HTML, and APIs. Experience deploying and debugging LLM systems, including containerization (Docker), API-based LLM integration (e.g., Ollama or vLLM), and environment configuration

Preferred Skills Background with graph-enhanced retrieval, using tools like Neo4j or ArangoDB, and an understanding of when and how to apply knowledge graphs to improve LLM grounding. Versed in model adaptation techniques, including LoRA, QLoRA, or PEFT approaches for fine-tuning or personalization. Expert in designing and implementing advanced prompt optimization frameworks, including developing automated evaluation systems and troubleshooting complex failure modes to enhance AI model performance and reliability.

Proven ability to design end-to-end hybrid search and reranking pipelines, such as ColBERT, BGE rerankers, or commercial tools like Cohere Rerank. Expertise with infrastructure optimizations, such as autoscaling (KEDA, HPA), Redis caching layers, or efficient streaming and batching. Demonstrated skill in safe deployment practices, including prompt injection mitigation and handling of sensitive or regulated data.

Clearance: Must be able to obtain/maintain a Secret clearance. Prefer holds an active Secret clearance. DUTIES & RESPONSIBILITIES Design and implement end-to-end RAG architectures, including document ingestion, chunking, embedding generation, vector indexing, query planning, retrieval, and response synthesis.

Evaluate and integrate LLMs, embedding models, and vector databases to support efficient and accurate retrieval and generation. Design and implement scaffolding and orchestration around LLMs, including prompt templating, tool invocation, evaluation harnesses, and safety guards. Develop data processing pipelines for structured and unstructured content (PDF, DOCX, HTML, Markdown, databases, APIs); implement normalization, deduplication, PII redaction, and metadata enrichment.

Implement and optimize retrieval strategies and context construction (citation, source attribution, grounding). Adapt retrieval and embedding strategies to domain-specific taxonomies, ontologies, or structured schemas; support contextual retrieval from hierarchical or relational sources. Productionize LLM-based systems: containerize components (Docker), deploy orchestration via Kubernetes or serverless platforms, implement observability (OpenTelemetry, logging, tracing), and manage configuration.

Measure and improve quality: define offline and online evals, golden datasets, A/B tests, hallucination detection, toxicity filters, and guardrails. Optimize performance and cost: batching, caching, streaming, and efficient context management. Implement security, privacy, and compliance best practices including access controls, injection defense, and safe data handling.

Develop solutions that can run entirely on-premise or in air-gapped environments, prioritizing data sovereignty and privacy. Various other duties in direct support of accomplishment of primary duties listed. SUPERVISORY/MANAGEMENT RESPONSIBILITY None.