Job Overview:
Pay Range: $55/hr - $60/hr
Requirement/Must Have:
- Expertise in LangChain, LlamaIndex, Semantic Kernel, AutoGen, or equivalent.
- Strong experience with Unity Catalog, Delta Lake, Vector Search, Databricks Workflows, and Model Serving.
- Hands-on with Lakehouse architecture patterns.
- Experience with open-source LLMs (Llama, Mistral, etc.), prompting techniques, and fine-tuning approaches.
- Strong knowledge of RAG architectures and embedding strategies.
- Expertise in LangChain, LlamaIndex, Semantic Kernel, AutoGen, or equivalent.
- Experience building agentic workflows and multi-agent systems.
- Advanced Python proficiency (APIs, web apps, orchestration, data processing).
- Familiarity with REST APIs and microservices architecture.
- Experience with MLflow, CI/CD pipelines, model lifecycle management, and observability tools.
- Knowledge of drift detection and model performance monitoring.
- Experience with Spark, SQL, and large-scale data processing.
- Familiarity with streaming frameworks (Kafka, Structured Streaming).
- Expertise in AI security risks (prompt injection, jailbreaks, data leakage).
- Experience implementing governance frameworks and compliance controls.
Responsibilities:
- Build and orchestrate autonomous AI agents with multi-step reasoning, tool usage, and workflow chaining using frameworks like LangChain, CrewAI, AutoGen, Semantic Kernel, or LlamaIndex.
- Deploy, fine-tune, and serve open-source LLMs (e.g., Llama 3) using Databricks Model Serving; optimize latency, throughput, and cost.
- Design advanced RAG pipelines leveraging vector search, embeddings, semantic ranking, and enterprise data sources (structured + unstructured).
- Develop prompt strategies, memory frameworks, and metadata tagging to improve contextual accuracy and response quality.
- Build intuitive AI-driven applications using Databricks Apps (Streamlit/Dash) or modern web frameworks to enable business consumption.
- Build reliable data pipelines (batch & streaming) supporting training, inference, and feature generation using Delta Lake.
- Implement enterprise-grade controls using Unity Catalog (row/column-level security, lineage, auditability) aligned with compliance standards.
- Implement guardrails (e.g., NeMo Guardrails) for prompt injection prevention, hallucination mitigation, and safe output handling.
- Establish CI/CD pipelines for AI models and agents, including versioning, monitoring, drift detection, observability, and incident response.
- Optimize model performance, GPU/compute usage, and inference cost efficiency across environments.
- Testing & Evaluation.
- Collaboration & Stakeholder Engagement.
- Documentation & Knowledge Transfer.
Founded in 2010 and headquartered in the Washington, DC metro area, Cynet Systems Inc. is a leading staffing and recruiting powerhouse. Proudly recognized as a nationally and locally certified diversity firm, Cynet delivers agile, scalable talent solutions across industries. With an active footprint in all 50 U.S. states and Canada, we support thousands of consultants through our expansive, high-performing recruitment engine operating across North America and Asia—ensuring speed, quality, and consistency in every hire.