Product ML Engineer: Build Memory-Driven AI Agents

Alpaca Research; Cambridge (MA); On-site; USD 140,000 - 200,000; Full time

Job summary

Alpaca Research in Cambridge, MA is seeking an ML engineer to bridge the gap between model capabilities and a complete system. You will work on model optimization, agent architectures, memory systems, and evaluation to make AI more capable, efficient, and reliable in local hardware.

The role combines ML research with building production-ready systems, including fine-tuning models, designing agent harnesses, and exploring new approaches to managing context and memory for long-running tasks.

Qualifications

• Experience with modern language models, transformers, PyTorch and the training/inference techniques behind them.
• Experience building or experimenting with LLM agents, tool calling, planning, retrieval, or multi-step reasoning systems.
• Understanding long-context processing and memory architectures within constrained context windows.
• Familiarity with model optimization, quantization, efficient inference, latency and compute tradeoffs.
• Comfort with reading papers, prototyping ideas, and delivering production-ready software.

Responsibilities

• Improve models: fine-tune, distill, and optimize language models for tool use and complex tasks.
• Build agent systems: develop harnesses, tool-use architectures, planning and execution loops.
• Engineer memory and context: design retrieval and persistent memory systems for long tasks.
• Optimize inference: explore quantization and inference-time techniques for local hardware.
• Evaluate and iterate: develop benchmarks for agent capabilities and real-world task performance.
• Research and implement: prototype new ideas and integrate successful approaches into production.

Skills

Modern LLMs

Transformers

PyTorch

Agent systems

Tools

Memory systems

Quantization

Job description

Alpaca Research in Cambridge, MA is seeking an ML engineer to bridge the gap between model capabilities and a complete system. You will work on model optimization, agent architectures, memory systems, and evaluation to make AI more capable, efficient, and reliable in local hardware.

The role combines ML research with building production-ready systems, including fine-tuning models, designing agent harnesses, and exploring new approaches to managing context and memory for long-running tasks.