Senior AI Application Engineer - LLM & VLM Enablement
Axelera AI3 hours ago
Remote, United Kingdom +38 moreSenior
Responsibilities
- Study new LLM and VLM architectures, prototype and profile them on Metis and Europa, and determine how they should run on Axelera hardware.
- Turn research results into supported Voyager SDK features with stable APIs, tests, documentation, and release gates.
- Own PyTorch reference implementations, quantization, weight mapping, and model-side contracts for LLM and VLM building blocks.
- Integrate prefill, decode, KV-cache management, streaming text output, and multimodal inputs into the Voyager SDK pipeline builder.
- Measure model accuracy against floating-point references and investigate or resolve quantization and implementation gaps.
- Extend the LLM/VLM bring-up platform with compiler, runtime, hardware, closed-loop, and evaluation capabilities.
- Recommend Voyager SDK capabilities and model-zoo demonstrations for GenAI, agents, and speech on edge devices.
Requirements
- At least 5 years of software engineering experience, including at least 2 years in ML inference, ML systems, or ML tooling.
- Strong understanding of transformer inference, including attention variants, KV caching, prefill versus decode, sampling, tokenization, and VLM architecture.
- Working knowledge of LLM quantization, including weight-only and weight-activation schemes, calibration, and outlier handling.
- Strong Python and PyTorch skills and experience shipping production-quality software.
- Daily experience using AI coding agents on real codebases, with judgment to verify or reject generated output.
- Ability to turn research code and hardware constraints into practical designs.
- Preferred: experience with inference engines or serving stacks such as llama.cpp, vLLM, ONNX Runtime GenAI, or ExecuTorch.
- Preferred: kernel development, tensor compiler experience, or profiling models on accelerator hardware.
- Preferred: experience with ASR, TTS, embedded-device agentic applications, ML or AI-agent evaluation harnesses, or tool-calling agents such as MCP servers.
Benefits
- Flexible working arrangement with office, fully remote, and relocation options across listed European locations.
- Attractive compensation package including a pension plan and extensive employee insurances.
- Option to receive company shares.
- Open, collaborative, inclusive culture with freedom and responsibility.
Categories
About Axelera AI
Axelera AI builds AI accelerators and an edge inference software stack (the Metis AI Platform) for computer vision and other workloads, selling chips, cards, and developer kits to OEMs and solution providers. Privately held and founded in 2021, it is headquartered in Eindhoven with R&D offices in Belgium, Switzerland, the UK, and Italy. The company focuses on on-device AI deployments where latency, power, and cost are critical.