SUMMARY
MARA is redefining the future of sovereign, energy-aware AI infrastructure. We’re building a modular platform that unifies IaaS, PaaS, and SaaS which will enable governments, enterprises, and AI innovators to deploy, scale, and govern workloads across data centers, edge environments, and sovereign clouds.
MARA is seeking a Lead Software Engineer to design, build, and scale systems that power agentic and intelligent workloads across our product ecosystem. This role blends deep expertise in machine learning application engineering, prompt orchestration, and retrieval-augmented generation (RAG) with strong software craftsmanship and automation discipline.
You will lead development of production-grade ML integrations—from model selection and evaluation to deployment pipelines, guardrails, and orchestration frameworks—ensuring that agentic systems are secure, reliable, and explainable. The ideal candidate thrives at the intersection of ML infrastructure, applied AI, and modern software engineering.
ESSENTIAL DUTIES AND RESPONSIBILITIES
- Lead architecture and development of agentic platforms that integrate multiple models, tools, and knowledge sources into dynamic reasoning systems.
- Evaluate and deploy foundation and open-source models (LLMs, vision, multimodal) using efficient inference strategies and fine-tuning where applicable.
- Design and maintain prompt lifecycle pipelines with version control, testing, and CI/CD integration (“PromptOps”).
- Build and optimize RAG systems—vector database configuration, retriever-generator orchestration, and embedding quality improvement.
- Implement guardrail frameworks for content safety, hallucination control, and policy enforcement across agentic workflows.
- Integrate and extend agentic frameworks (LangChain, LangGraph, CrewAI, AutoGen, or equivalent), both in code-based and visual orchestration environments.
- Collaborate with data, product, and infrastructure teams to design scalable APIs and services that enable model-driven applications.
- Define observability and evaluation metrics for model performance, latency, and behavior drift in production.
- Drive best practices for secure AI development, privacy-preserving data handling, and governance of third-party model integrations.
- Mentor engineers across ML, backend, and platform domains; champion continuous learning and experimentation.
QUALIFICATIONS
- 8+ years of professional software engineering experience, including 3+ years in ML application development or AI platform engineering.
- Proficiency in Python, with strong understanding of ML toolchains (PyTorch, Hugging Face, LangChain, MLflow, Ray, etc.).
- Proven experience with model evaluation, fine-tuning, and deployment across cloud and on-prem environments.
- Hands-on experience with RAG architectures and vector databases (Weaviate, Milvus, pgvector, LanceDB, FAISS).
- Deep understanding of prompt design, orchestration, and versioning using CI/CD workflows and automated testing frameworks.
- Familiarity with agentic systems, both code-driven and visual-builder interfaces (LangGraph Studio, Dust, Flowise, Relevance AI, etc.).
- Strong knowledge of guardrail techniques (rule-based filters, policy evaluators, toxicity detection, grounding validation).
- Experience deploying ML systems on Kubernetes and serverless environments with observability (Prometheus, Grafana, OpenTelemetry).
- Solid understanding of API design, microservice architecture, and data pipeline integration.
- Excellent communication and leadership skills, with ability to translate complex ML concepts into actionable engineering outcomes.
PREFERRED EXPERIENCE
- Background in HPC, ML infrastructure, or sovereign/regulated environments.
- Familiarity with energy-aware computing, modular data centers, or ESG-driven infrastructure design.
- Experience collaborating with European and global engineering partners.
- Strong communicator who can bridge engineering, business, and vendor ecosystems seamlessly.
United States
