AI Engineer — Validation & Inference Runtime

Overview

**Work on real-time inference, validation, and production voice AI latency.

Krim is building KrimOS — the autonomous agent operating system for banking.

We are hiring an AI engineer to make AI systems trustworthy in real-time voice.

At the core of KrimOS is KendraOS — a pre-execution validation layer that checks AI actions before they execute. Your role is to make this inference and validation stack fast enough for production voice AI.

What you’ll work on:

• Inference runtime optimisation
• Validation execution engine design
• Real-time voice AI latency reduction
• Parallel validator execution
• Model routing and serving
• Production monitoring for latency and validation performance

Relevant stack includes:

• vLLM
• TensorRT-LLM
• SGLang
• AWQ / GPTQ / FP8
• Continuous batching
• Speculative decoding
• Real-time inference systems

What we’re looking for:

• 1–3 years of relevant experience
• Voice AI or real-time inference experience preferred
• Strong Python and systems programming mindset
• Experience shipping ML systems to production
• Familiarity with open-source model serving frameworks
• Strong interest in inference optimisation and latency engineering
• AI coding tools as part of daily workflow

Why join:

• Work on a difficult, differentiated problem
• Direct founder access
• High ownership very early
• Meaningful equity
• Real production deployments in banking environments

To apply, email nath@krim.ai with subject:
“Inference Engineer — [Your Name]”

Please include:

• Brief note or short video on what you’ve built
• Your AI coding workflow
• GitHub or public inference-related work