ML Research Intern in Agentic Runtime Systems
Core
Researching agentic runtime systems to build reliable multi-step task execution by treating LLMs as components within deterministic control systems.
Role type
ML Research Intern (Agentic Runtime Systems)
Builds
Agentic Harness, verification layers, constrained decoding systems, KV-cache optimization strategies
Domain
Artificial Intelligence, Agentic Systems, Large Language Model Engineering
Deliverable
production ML models
Required skills
Systems thinking, Machine Learning, Python, Rust, State machine design, Constrained decoding, KV-cache optimization, Error accumulation handling, Self-correction mechanisms, Context drift analysis
Preferred skills
Experience with deterministic guardrails, Dynamic tool loading optimization, Raw model output handling, Software robustness engineering
Technologies
Python, Rust, LLMs, KV-cache
Responsibilities
Architect loops that survive error accumulation, Implement verification layers for self-correction, Design restorable compression strategies to prevent attention decay, Optimize KV-cache hit rates, Build systems where agents compile and enforce constraints, Analyze agent traces to identify context drift
