Senior ML Inference Platform Engineer

About AION: AION is building the next generation of AI cloud platform by transforming the future of high-performance computing (HPC) through its decentralized AI cloud. Purpose-built for bare-metal performance, AION democratizes access to compute power for AI training, fine-tuning, inference, data labeling, and full stack AI/ML lifecycle. Led by high-pedigree founders with previous exits, AION is well-funded by major VCs with strategic global partnerships. Headquartered in the US with global presence, the company is building its initial core team across India, London and Seattle. Who You Are : You're an ML systems engineer who's passionate about building high-performance inference infrastructure. You don't need to be an expert in everything - this field is evolving too rapidly for that - but you have strong fundamentals and the curiosity to dive deep into optimization challenges. You thrive in early-stage environments where you'll learn cutting-edge techniques while building production systems. You think systematically about performance bottlenecks and are excited to push the boundaries of what's possible in AI infrastructure. Requirements Key Responsibilities : Build and optimize LLM inference systems working towards 2-4x performance improvements over standard frameworks like vLLM and TensorRT-LLM Implement modern inference optimizations including KV-cache management, dynamic batching, speculative decoding, compression and quantization strategies Develop GPU optimization solutions using CUDA, with opportunities to learn advanced techniques like Triton kernel development and CUDA graphs Design model evaluation and benchmarking systems to assess performance across reasoning, coding, and safety metrics Research and integrate trending open-source models (DeepSeek R1, Qwen 3, Llama 4, Mistral variants) with optimized configurations Build performance monitoring and profiling tools for GPU cluster analysis, bottleneck identification, and cost optimization Create cost-performance optimization strategies that balance throughput, latency, and infrastructure costs Explore agent orchestration capabilities for multi-step reasoning and tool integration workflows Collaborate with tech and product teams to identify optimization opportunities and translate them into production improvements Skills & Experience : High agency individual looking to own and influence product architecture and company direction 3+ years of software engineering experience with focus on performance-critical systems and production deployments Strong Python expertise and working knowledge of C++ for performance optimization Working understanding of deep learning fundamentals including transformer architectures, attention mechanisms, and neural network training/inference Hands-on experience of model serving and deployment techniques Experience with at least one modern inference framework (vLLM, TensorRT-LLM, SGLang or similar) in a production setting Hands-on experience with PyTorch including model development, training loops, and basic distributed computing concepts Understanding of distributed systems concepts including load balancing, auto-scaling, and fault tolerance Basic GPU programming experience with CUDA or willingness to quickly learn GPU optimization techniques Strong debugging and performance profiling skills for identifying and resolving system bottlenecks Benefits : Join the ground floor of a mission-driven AI startup revolutionizing compute infrastructure Work with a high-caliber, globally distributed team backed by major VCs Competitive compensation and benefits Fast-paced, flexible work environment with room for ownership and impact Hybrid model: 3 days in-office, 2 days remote with flexibility to work remotely for part of the year

#J-18808-Ljbffr