TL;DR
Staff Software Engineer (AI): Building and optimizing distributed training platforms for LLM and multimodal fine-tuning and post-training, and owning inference architecture across multiple providers. Focus on designing and implementing state-of-the-art training algorithms, researching inference optimizations, and improving latency and cost efficiency across the ML stack.
Location: Hybrid in San Francisco
Salary: $300,000–$430,000 + Equity
Company
hirify.global is the leading conversational AI platform empowering brands to deliver concierge customer experiences through AI agents.
What you will do
- Design and build distributed training platforms for LLM and multimodal fine-tuning and post-training at scale.
- Implement and integrate state-of-the-art training algorithms into production pipelines.
- Own inference architecture and multi-provider routing, including failover and optimization.
- Research and implement inference optimizations including quantization, speculative decoding, and batching strategies.
- Lead initiatives to improve latency and cost efficiency across the training and serving stack.
- Build evaluation and experimentation infrastructure that enables rapid, reliable iteration.
Requirements
- 8+ years building ML infrastructure or production systems at scale.
- Deep experience with distributed training: multi-node GPU clusters, fault tolerance, and optimization.
- Strong understanding of LLM inference: latency optimization, provider tradeoffs, and serving architecture.
- Proficiency in Python and modern ML frameworks (PyTorch, JAX, or TensorFlow).
- Proven track record leading complex, multi-quarter technical projects.
Culture & Benefits
- Medical, dental, and vision benefits.
- Take what you need vacation policy.
- Daily lunches, dinners and snacks in the office.
- An in-office company driven by a shared commitment to excellence and velocity.
Будьте осторожны: если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого - это мошенники. Обязательно жмите "Пожаловаться" или пишите в поддержку. Подробнее в гайде →