🛠 Tech Stack

💼 About This Role

You'll optimize LLM inference performance for the Kimchi platform, continuously matching customer workloads to the most cost-efficient serving configuration. Your work directly improves margins and customer p99 latency.

🎯 What You'll Do

• Push continuous batching and speculative decoding throughput
• Profile and optimize TTFT and TPOT latency
• Maximize KV cache utilization with paged attention and caching
• Quantize weights and KV using INT8/INT4/FP8 without quality regression

📋 Requirements

• 5+ years building ML inference or training infrastructure
• Strong Python skills for production services
• Hands-on experience with vLLM, SGLang, or TensorRT-LLM
• Fluency in quantization tradeoffs with measured quality impact

✨ Nice to Have

• Experience with distributed inference topologies
• Knowledge of GPU memory accounting and kernel tuning

🎁 Benefits & Perks

• 💰 Competitive salary with equity options
• 🌍 Remote-first global environment
• 📚 Learning budget for conferences and courses
• ⚡ Fast-paced workflow with features in 1-4 weeks
• 🎮 Annual hackathon and team-building events

📨 Hiring Process

Estimated timeline: 2-4 weeks

1. 1Screening call with Recruiter· 30 min
2. 2Hiring Manager interview· 45 min
3. 3Technical interview (system design)· 60 min
4. 4Live coding· 60 min
5. 5Culture Check interview with executive· 45 min