AI/ML Engineer

Cyrano is a stealth, pre\-seed, revenue\-generating AI startup in San Francisco building systems with real memory and continuity. We're hiring an early AI/ML engineer who wants real ownership (not another big\-company role) and is ready to prove it. What you'll do \- Take ambiguous problems from idea to production across LLM features, agent workflows, retrieval, and memory \- Build async Python services, including streaming and real\-time interactions \- Design short\-term, long\-term, and episodic memory for persistent conversational state \- Build evaluation datasets, catch regressions, and improve quality, latency, and cost \- Document decisions and own your work from design through delivery What you'll bring \- Strong Python (asyncio, FastAPI) plus working knowledge of C\+\+, Rust, or Go \- Solid CS foundations: graphs, relational theory, algorithmic complexity \- Hands\-on experience shipping LLM features: RAG, tool use, agent frameworks (LangChain, AutoGen, CrewAI, LlamaIndex, or custom) \- Practical deep learning: transformers and PyTorch, JAX, or TensorFlow \- Fine\-tuning / post\-training (SFT, LoRA/QLoRA, preference optimization), and the judgment to know when not to \- Production infrastructure: containers, Kubernetes, cloud, CI/CD, observability Nice to have: real\-time voice/speech, LLM serving (vLLM, TensorRT\-LLM, Triton), distributed training, open\-source work or shipped AI products. We care about what you've actually built more than a list of tools. A CS/ML degree is useful but not required. How we hire (about one week) 1\. Self\-assessment form (\~30 min) 2\. JokeBox Challenge (invite only): opens Mon, Sep 28; Box due Tue, Sep 29; full build due Wed, Sep 30 3\. Finalist Hackathon: all\-day, in person in SF, Thu, Oct 1, with on\-the\-spot offers Full\-time · In person, SF Bay Area (Bay Area residence required) · $120K–$200K base \+ early equity Pay: $120,000\.00 \- $200,000\.00 per year Work Location: Hybrid remote in San Francisco, CA 94158