Senior AI & LLM Systems Engineer

Syscov; United States; Remote; USD 140,000 - 220,000; Full time

Job summary

Syscov is hiring a senior AI/ML engineer to design custom model routing engines and autonomous multi-agent systems. This remote, full-time role targets experienced professionals with a focus on scalable LLM backends and high-throughput inference for global deployments.

Ideal candidates have strong Python systems programming, CUDA/PyTorch memory management, and hands-on experience with vector databases and contextual retrieval systems.

Qualifications

• 3+ years of AI/ML engineering experience with production exposure to LLMs and embedding architectures.

Responsibilities

• Develop custom LLM orchestration backends with multi-model failover and speculative routing.
• Build high-precision RAG pipelines using hybrid semantic/dense embeddings and reranking.

Skills

AI/ML engineering

Python systems programming

CUDA memory management

Vector databases

Inference throughput optimization

Job description

Design custom model routing engines, high-throughput inference gateways, and autonomous multi-agent systems.

Location Remote (Global / US / Pakistan)

Level Senior

Experience 3+ Years

Type Full-Time

Stack

What you will do

• Develop custom LLM orchestration backends with intelligent multi-model failover and speculative routing.
• Build high-precision RAG pipelines using hybrid semantic/dense embeddings and reranking algorithms.
• Deploy self-hosted open weights models on GPU clusters with vLLM, TensorRT-LLM, and quantization (AWQ/GPTQ).
• Instrument rigorous model evaluation benchmarks (evals) for hallucination detection and response latency.

What we are looking for

• 3+ years of AI/ML engineering experience with direct production exposure to LLMs and embedding architectures.
• Strong Python systems programming skills with FastAPI, asynchronous IO, and CUDA/PyTorch memory management.
• Deep understanding of vector databases, contextual retrieval mechanics, and agentic workflow frameworks.
• Experience optimizing inference throughput (tokens/sec) and GPU utilization.