Sobhan Bahrami

Meet Sobhan

1 following6 followers

Forward Deployed AI Engineer and Distributed Systems Architect specializing in embedding into high-stakes environments to design, build, and deploy production AI pipelines from zero to scale.

Where pure research teams stop at the notebook and traditional software teams struggle with model internals, I bridge the gap: training specialized models from scratch, architecting low-latency distributed runtimes, and shipping resilient, compliant enterprise systems.

🎯 What I Deploy in Production:
• Forward Deployed AI & Solutions: Custom enterprise agentic architectures, substrate runtimes, RAG/KAG, human-in-the-loop governance, and real-time voice systems.
• Distributed ML Infrastructure: PyTorch DDP/FSDP, DeepSpeed ZeRO, Ray Train/Core, vLLM, Triton, and high-throughput multi-node clusters.
• Privacy-Preserving & Federated AI: Flower (flwr), NVIDIA FLARE, Differential Privacy (DP-SGD), secure aggregation for decentralized client networks.
• High-Performance Engineering: Rust, Go, Python (AsyncIO), WebGPU/WGSL, and sub-10ms event-driven trading/state engines.

🏆 Track Record:
• 100+ production systems delivered globally; 99.99% uptime for 100K+ concurrent users.
• MSc in AI Systems (UFAR) | IBM Distributed Systems Project | 2x IEEE/ACM Code Go Gold Medalist.
• Author: MoEDCoder (Diffusion MoE Research) & The Symbiote Architecture.

Available for Forward Deployed AI Lead roles, enterprise AI infrastructure builds, and fractional solutions architecture contracts.