What you'll do
- Lead development of scalable training pipelines for large transformer and LLM models, covering data ingestion to distributed training and evaluation.
- Build and optimize low-latency, high-reliability inference services with features like autoscaling, routing, and fallbacks.
- Work on GPU-level performance tuning, including kernel optimization and utilization improvements across training and inference stacks.
- Collaborate closely with ML scientists to implement and productionize cutting-edge AI training and inference methods.
- Contribute to team growth by mentoring engineers and raising technical standards for reliability and operational excellence.
What you should know
- Ideal candidates have 5+ years of software engineering experience with a strong record of shipping high-quality products.
- Applicants should be comfortable working on large-scale production environments involving significant traffic and data.
- The role offers exposure to state-of-the-art AI infrastructure, including GPU programming and transformer model training.
- Intercom supports continuous learning and development, encouraging engineers to invest in their own growth.
- The position requires strong communication skills to collaborate effectively with both technical and non-technical stakeholders.
About the company
- Intercom is a leading AI customer service company founded in 2011, trusted by nearly 30,000 global businesses.
- The company emphasizes a culture of pushing boundaries, speed, and delivering incredible customer value.
- Intercom maintains a radically open and accepting culture, focusing on inclusivity and avoiding divisive topics internally.
- They have a hybrid work policy requiring onsite presence at least three days a week to foster collaboration and culture.
- Intercom values diversity and equal opportunity, with strong policies against discrimination on all legally protected bases.
Key required skills
transformersLLMsCUDATritonPython