Senior Platform Machine Learning Engineer

Launch Legends•Cheyenne, WY

31d

About The Position

We are seeking an exceptional Senior Platform Machine Learning Engineer to join our pioneering team as an Equity Co-Founder at the forefront of native blockchain execution. At Autheo, we are building the platform that redefines decentralized computation, delivering sub-microsecond execution speeds, billions of transactions per second throughput, and mathematically proven security against adversaries. You will own sovereign model science for trillion-parameter GenAI, agentic systems, RAG/KAG retrieval, and predictive analytics — training across 10,000+ DePIN GPUs with federated learning, ε=0.5 differential privacy, zk-proof provenance, and 100× speedup while maintaining HIPAA/GDPR compliance on healthcare and DeFi data.

Requirements

8+ years shipping production ML models with PyTorch/TensorFlow at scale
Multiple federated learning deployments on 1,000+ heterogeneous nodes
Hands-on GNN (PyG/DGL) + KNN hybrid systems and GenAI fine-tuning for code/reasoning
Proven ε=0.5 differential privacy and/or zk-proof model verification in production
Deep expertise in distributed training (DeepSpeed, Horovod, Ray) and MLOps (Kubeflow, MLflow)

Nice To Haves

PhD + first-author papers at NeurIPS/ICML/ICLR
Experience training on healthcare (FHIR) or financial on-chain data
Patents in federated learning, privacy-preserving ML, or RAG/KAG

Responsibilities

Ship production federated training loops (Flower/FedML) for transformers, GNNs, and KNN models across global DePIN fleets
Fine-tune Llama/CodeLlama-class models generating 95 % compliant GrappLang code from natural-language prompts
Build hybrid GNN-KNN pipelines achieving 99 % precision on DeFi fraud and DePIN node-failure prediction
Design RAG embeddings and KAG graph reasoning systems delivering 99 % hallucination-free retrieval over exabyte sovereign data
Embed Opacus differential privacy (ε=0.5) and zk-proof model lineage into every training run
Build continual learning pipelines adapting to FHIR R5 updates and new DeFi protocols without full retraining
Deploy 1 M+ inferences/sec GenAI endpoints with 99.99 % uptime via Kubeflow/Seldon and DeepSpeed
Implement RLHF loops optimizing regulatory accuracy using enterprise reviewer feedback
Publish sovereign federated learning and zk-ML results at NeurIPS, ICML, ICLR
Keynote at NeurIPS 2026 on “Trillion-Parameter Federated GenAI with Provable Privacy”