LLM Systems Engineer | Distributed Inference Specialist | Performance Optimization Enthusiast

I build scalable, high-performance systems that push the boundaries of what's possible with large language models. My focus is on bridging the gap between cutting-edge ML research and production-grade infrastructure.

I design and implement distributed systems for efficient LLM inference, tackling challenges like:

• Sparse Expert Routing — Optimizing mixture-of-experts (MoE) models for inference
• Distributed Caching Strategies — Smart prefetching and cache management for latency reduction
• Request Scheduling & Batching — Maximizing throughput while maintaining response quality
• Model Quantization — Reducing model size without sacrificing accuracy
• Performance-First Architecture — Building systems that scale from edge devices to data centers

Building a production-grade inference engine that intelligently routes requests to specialized experts, manages distributed storage efficiently, and maintains low latency at scale.

Key Components:

• 🧠 Expert Routing: Intelligent predictor for selecting optimal experts
• 💾 Multi-tier Storage: Local SSD, network backends, and intelligent prefetching
• 🔄 Request Scheduling: Batch optimization with dynamic queue management
• ⚡ Performance Optimization: Quantization, caching, and pipeline prefetching

Languages: Python · TypeScript · JavaScript · Java · HTML5

ML & Systems:

Backend & DevOps:

Frontend:

Intelligent LLM inference engine with expert routing and smart caching

• Expert router with performance prediction
• Multi-tier storage (local SSD + network backends)
• Dynamic request scheduling and batching
• Quantization support for model compression
• Prefetching pipeline for latency reduction

Experience building end-to-end web applications with modern tech stacks, from backend APIs to responsive frontends

Optimize for impact, not perfection. The best system is one that ships and learns. I focus on identifying real bottlenecks, solving them elegantly, and measuring the results.

• Performance-First Mindset — Every millisecond matters in inference
• System Thinking — Understanding end-to-end data flow and trade-offs
• Production Readiness — Building for scale and reliability from day one
• Continuous Learning — Staying ahead in the rapidly evolving ML infrastructure space

Open to collaborations on distributed systems, ML infrastructure, and performance optimization projects.