About Us

Scroll down the page to see all associated job requirements, and any responsibilities successful candidates can expect.

Inception is a generative AI startup. Leveraging breakthrough AI research, we have developed a platform for training next-generation large language models (LLM) powered by diffusion. Unlike existing auto-regressive models, which only output one token at a time, diffusion LLMs can output many tokens in parallel. This means that they are several times faster and can leverage their additional test-time compute to improve quality. They also enable fine-grained control over their outputs to adhere to specific schema and semantic constraints, and they provide a unified paradigm for combining language with other data modalities, including audio, images, and videos.

Our team is led by Stefano Ermon (co-inventor of diffusion models, flash attention, and DPO; faculty at Stanford), Aditya Grover (co-inventor of node2vec and decision transformers; faculty at UCLA), and Volodymyr Kuleshov (prev. co-founder and CTO at Afresh Technologies; faculty at Cornell), and includes engineers from Google Deepmind, Meta AI, Microsoft AI, and OpenAI. We are currently deploying large-scale diffusion LLMs at Fortune 500 companies.

Role Overview

We are looking for Research Scientists / Engineers with deep expertise in training and optimizing large language models. In this role, you will work on advancing our diffusion-based LLM architecture, developing novel training techniques, and pushing the boundaries of what's possible with parallel token generation.

Key Responsibilities

Design, develop, and optimize LLM architectures and models. Implement innovative approaches for training, fine-tuning, and scaling generative AI models. Work on data preprocessing pipelines, model evaluation, and alignment to enterprise use cases. Design and implement novel model architectures for diffusion-based language models Develop and optimize training objectives and loss functions for LLMs Research and implement techniques for controlled text generation and constraint satisfaction Develop methods for multi-modal integration within the diffusion framework Work on improving model efficiency, reducing training time, and optimizing inference

Qualifications

BS/MS/PhD in Computer Science or a related field (or equivalent experience). At least 2 years of experience working on ML projects in PyTorch (or equivalent DL framework), preferably in a research lab or engineering role. Excellent familiarity with transformers and fundamental LLM concepts (e.g., autoregressive pretraining, instruction tuning, in-context learning, and KV caching). Familiarity with training and inference in diffusion models. Experience with training deep learning models at scale using distributed computing environments.

Preferred Skills

Extensive experience training transformer-based language models from scratch Knowledge of advanced training techniques (mixed precision, gradient accumulation, etc.) Experience with multi-modal learning and cross-modal architectures Background in optimization theory and neural network architecture design Experience with LLMs serving frameworks like vLLM, SGLang, or TensorRT.

Why Join Us

Impact: Deploy LLMs that transform how millions of users work, create, and solve real-world problems. Innovation: Pioneer novel architectures and training techniques for diffusion LLMs. Growth: Enjoy a fast-paced, collaborative environment where your contributions will directly shape the future of generative AI.

Perks & Benefits

Competitive salary and equity in a rapidly growing startup. Flexible vacation and paid time off (PTO). Health, dental, and vision insurance. Professional development opportunities (conferences, courses, etc.).

This is an exciting opportunity to join a startup at the forefront of AI development! If you’re ready to make a tangible impact in the world of generative AI, apply today.

We are an equal opportunity employer and encourage candidates of all backgrounds to apply.

PI274778171 #J-18808-Ljbffr