About Liquid AI

Spun out of MIT CSAIL, we build general-purpose AI systems that run efficiently across deployment targets, from data center accelerators to on-device hardware, ensuring low latency, minimal memory usage, privacy, and reliability. We partner with enterprises across consumer electronics, automotive, life sciences, and financial services. We are scaling rapidly and need exceptional people to help us get there.

The Opportunity

Developers do not yet grasp what a fine-tuned small model can do until they see one running on a phone, in a browser, or on a Jetson board. Closing that imagination gap is the job. You will be Liquid's technical voice where builders already are (Hugging Face, GitHub, Discord, and the developer events that matter for edge AI), turning individual troubleshooting sessions on chat templates, quantization, and on-device deployment into cookbooks, reference apps, and content the whole community can use.

This is a hands-on role within Marketing & Communications. You will work closely with our model, platform, and product teams: they partner on technical accuracy, and you own how the work is explained, presented, and adapted for developers.

What We're Looking For

We need someone who:

Builder: You would rather ship a working reference app than write a strategy memo, and you measure yourself by what developers actually adopt.

Community native: You answer hard questions in public, you are comfortable being wrong in front of people, and you know the small-model and edge-AI ecosystem well enough to hold real opinions.

Trusted technical voice: You move between a quick Discord answer, a step-by-step cookbook, and a long-form technical post without losing accuracy or voice, and you earn the trust of skeptical developers by staying concrete and avoiding hype.

In-person energizer: Hackathons, workshops, hallway-track demos, and the conferences where builders actually show up are where you do your best work.

End-to-end owner: You take ownership in ambiguity and turn one-off problems into reusable resources without being asked.

The Work

Be the technical voice in the community: live where LFM developers already are (Hugging Face, GitHub, Discord, X), answer hard questions in public, maintain Liquid's presence on the Hub, and amplify community fine-tunes and edge deployments worth seeing.

Show up in person: host hackathons, build nights, and technical workshops at Liquid offices and partner venues; submit talks, run booths, and demo at the conferences that matter for small-model and edge AI; and partner with hackathon organizers to make LFMs the obvious choice for builders who care about latency, on-device, or cost.

Build the on-device and adaptation story: ship reference applications on real edge hardware (iOS, Android, browser, Jetson, NPUs), maintain integration recipes for llama.cpp, MLX, ONNX, ExecuTorch, and write cookbooks that take a developer from base model to fine-tuned, quantized, and deployed in one sitting.

Create content that earns trust: deeply technical blog posts, demos, and honest benchmarks against Qwen, Gemma, Phi, and SmolLM, and turn the best workshop material into evergreen content so the in-room audience compounds online.

Close the loop: keep a friction log of developer pain points (chat templates, tokenizers, deployment gotchas) and bring that signal into roadmap conversations with the model and platform teams.

Experience & Capabilities

What you'll need

Hands-on LLM experience, including fine-tuning (LoRA, QLoRA, full fine-tuning, distillation) and tokenizer debugging, with a track record of shipping to production or active community environments.

Fluency with PyTorch, Hugging Face (Transformers, PEFT, Datasets), and at least one model serving framework (llama.cpp, MLX, vLLM, ONNX Runtime, or TGI), plus a working understanding of quantization tradeoffs (GGUF, AWQ, GPTQ, INT8/INT4).

Experience with on-device deployment (iOS, Android, embedded) or specialized work in the efficient-model ecosystem (Phi, Gemma, Qwen, SmolLM, or distilled architectures).