Evaluate and select optimal model architectures (LLMs, SLMs, or traditional ML) based on mission requirements, considering tradeoffs between accuracy, latency, and cost.
Guide customers on "Build vs. Buy vs. Fine-tune" decisions, prioritizing open-source models (Llama, Mistral, Falcon) that can run securely within a sovereign data perimeter.
Experience building Agentic Workflows (AI agents that can execute API calls and multi-step tasks).
Design and implement robust data pipelines within CDP to transform "messy" legacy data into AI-ready formats.
Develop and optimize Vector Databases and Retrieval-Augmented Generation (RAG) architectures to ground AI responses in verified agency facts.
Build Data pipelines with Spark, Nifi, Kafka or other ETL tools.
Optimize model inference for production environments using quantization, pruning, and hardware acceleration (NVIDIA GPU orchestration).
Implement LLMOps to monitor model performance, detect hallucination rates, and manage model versioning and drift.
Collaborate with the customer’s AI Center of Excellence (CoE) to establish automated guardrails for ethics, bias mitigation, and FedRAMP/IL5 compliance.
Translate complex technical AI concepts into mission-value briefings for GS-level stakeholders and agency leadership.
Requirements
5+ years in Data Engineering, Machine Learning, or Software Engineering, with at least 2 years focused on Generative AI or Deep Learning.
Expertise in Python and deep learning frameworks (PyTorch, TensorFlow, Hugging Face).
Hands-on experience with Cloudera (CDP), Spark, or similar big data ecosystems.
Proficiency in orchestration tools like LangChain, LlamaIndex, or Haystack.
Experience developing visual data representations and dashboards (Django, React, or Angular).
Experience using a compiled programming language, preferably one that runs on the JVM (Java, Scala, etc).
Proven ability to build ETL/ELT pipelines and work with both SQL and NoSQL/Vector databases (e.g., Pinecone, Milvus, or PGVector).
Understanding of government security frameworks (NIST AI RMF, FedRAMP, SRGs, STIGs).