over 1 year ago
Responsibilities
- Build secure, scalable, automated pipelines for LLM fine-tuning, supervised fine-tuning, LoRA, RLHF, DPO, RAG embeddings, model conversion, quantization, and inference rollout.
- Manage hybrid cloud, on-premises, and GPU-cluster infrastructure for training and inference using Kubernetes, Ray, and Terraform.
- Containerize models and agents with Docker and operate reproducible CI/CD workflows using GitHub Actions or ArgoCD.
- Implement model governance covering versioning, metadata, lineage, reproducibility, and evaluation capture.
- Create evaluation and benchmarking frameworks using tools such as OpenLLM-Evals, RAGAS, and LangSmith.
- Integrate policy, security, and access-control layers including OPA, ABAC, and Keycloak for tenant-specific model policies.
- Instrument observability for latency, token usage, performance, errors, tracing, and model drift.
- Support agentic application deployments using LangGraph, LangChain, vLLM, TGI, and Triton.
- Own production systems end to end and contribute to architecture, performance, cost optimization, and deployment workflows.
Requirements
- 4+ years of experience in MLOps, ML platform engineering, or infrastructure-focused ML roles.
- 5+ years as a full-stack or backend engineer with experience owning and delivering production systems end to end.
- Deep familiarity with MLflow, Weights & Biases, DVC, and HuggingFace Hub.
- Experience deploying large open-source models such as LLaMA, Mistral, Falcon, or Mixtral.
- Experience with HuggingFace Trainer, DeepSpeed, FSDP, or QLoRA.
- Experience with inference serving using vLLM, TGI, Ray Serve, or Triton Inference Server.
- Proficiency with Terraform, Helm, Kubernetes, container orchestration, and CI/CD for ML.
- Experience managing hybrid GPU workloads using services or platforms such as Lambda, Modal, HuggingFace Inference, or SageMaker.
- Familiarity with LangChain, LangGraph, LlamaIndex, RAG orchestration, embedding pipelines, and vector databases.
- Experience with model-level RBAC, usage tracking, audit trails, API rate limits, tenant billing, SLA observability, OPA, and Rego.
- Experience with modern frontend frameworks such as React or Next.js and familiarity with APIs, databases, cloud infrastructure, or deployment workflows.
- Strong ownership, communication, architecture, performance, iteration, and startup collaboration skills.
- Preferred experience includes founder, founding engineer, or 0-to-1 pre-seed startup work and familiarity with SOC 2, HIPAA, or GovCloud-grade operations.
Benefits
- Full-time founding-team role in the San Francisco Bay Area.
- Competitive salary and meaningful founding-tier equity.
- High autonomy and ownership in an early-stage startup environment.
- Opportunity to work across AI research, DevOps, backend engineering, open-source LLMs, agentic systems, and production infrastructure.
Tech Stack
Categories
About Fabrion
Fabrion builds enterprise AI infrastructure centered on agent-based systems, knowledge graphs, and multi-tenant governance for mission-critical workflows. The privately held company operates an Agentic AI Lab to train and evaluate specialized models, offering software and platform services to large organizations; founded in 2025 and headquartered in San Francisco.