Sr. Research Scientist

Open Problems You Could Own in Your First Year

• Long-video understanding. Reasoning over multi-hour egocentric footage where the events that matter are sparse. This requires temporal grounding, long-context modeling, and memory beyond what current VLMs offer.
• Post-training for expert-level perception. Using SFT and RL (e.g., GRPO) to push VLM labeling of fine-grained construction activity to parity with expert human taggers, across every trade.
• Inference at fleet scale. We will soon collect over 10,000 hours of new video per day, expanding quickly from there. Making frontier-quality inference cheap enough to run on all of it, through distillation, quantization, and model routing, is partially unsolved.
• Evals that predict reality. Building benchmarks that track real field performance, not just standard academic metrics.

• Design, train, and iterate on vision-language models fine-tuned for spatial intelligence in construction environments. Model quality is the single most important output of this role.
• Own the model training and evaluation pipelines end to end, so we can rapidly experiment, measure performance, and deploy models into production.
• Push the frontier of what's possible on our data, from establishing baselines with state-of-the-art models to developing novel post-training recipes, long-context architectures, and visual reasoning techniques.
• Build and expand our benchmark suite across video question answering, temporal reasoning, activity recognition, and site-level analytical reasoning.
• Design evaluation metrics that measure real-world construction task performance, not just standard academic benchmarks.
• Make our benchmark suite the reference point that future construction AI research measures against.
• Apply distillation, quantization, and model routing so state-of-the-art understanding runs affordably across thousands of hours of daily footage.
• Partner with our hardware and infrastructure teams on system design so research decisions and deployment realities inform each other.
• Own the end-to-end story from research paper to production model running on active jobsites.
• Work directly with the Chief Science Officer to define the research priorities that matter most for Ironsite's next twelve months and next three years.
• Contribute to the intellectual culture of the team through paper reading, technical mentorship, and setting the bar for how we do research at Ironsite.

• Training large-scale models efficiently under real compute budgets while maximizing performance on the problems that matter for our customers.
• Developing novel pre-training and post-training objectives that capture construction-specific knowledge, temporal reasoning, and fine-grained perception.
• Implementing efficient attention mechanisms and architectural innovations for long-context understanding of construction workflows that span hours or days.
• Designing evaluation metrics that measure real-world construction task performance beyond standard benchmarks, and holding the whole team accountable to what those metrics actually predict.
• Balancing model capability with deployment constraints for edge, on-prem, and cloud inference across the fleet.

• 5+ years of hands-on experience designing and training large-scale deep learning models, particularly transformer-based architectures, with at least 2 spent operating at a senior research level at a company or lab you're proud of.
• Deep expertise with modern deep learning frameworks (PyTorch, JAX, or similar) and strong proficiency in Python with solid software engineering fundamentals.
• Experience working with and creating large-scale vision or language datasets.
• Comfort operating at the edge of what's known. You've worked on research problems where the right answer wasn't in a paper yet, and you've figured it out anyway.
• A background in Computer Science, Machine Learning, AI, Robotics, or a related field.
• Deep expertise in fine-tuning and post-training large language or vision-language models (SFT, GRPO and other RL methods, parameter-efficient tuning such as LoRA).
• Hands-on experience with the challenges of video data, including temporal reasoning, long-context modeling, and efficient processing.
• Experience optimizing inference at scale, including quantization, distillation, sparsity, and efficient serving.
• Track record of publications in top-tier AI, ML, or CV conferences.

• Foundational impact. Solve fundamental AI problems to transform one of the world's largest and least-digitized industries. Your models ship to real jobsites, not just papers.
• Ownership and autonomy. We're a fast-paced startup where you'll have significant ownership over core research directions. We value intellectual curiosity, first-principles thinking, and iterating quickly to turn ambitious ideas into reality.
• Dream dataset. Exclusive access to a massive, proprietary, and continuously growing corpus of egocentric jobsite video from hundreds of devices deployed on active construction sites. A moat that enables frontier research.
• World-class team. Collaborate with a small, elite team of researchers and engineers who have shipped cutting-edge AI products at companies like DeepMind, Etched, Meta, Apple, and NVIDIA.

• Base salary: $200k-$400k per year, commensurate with experience
• Significant early-stage equity
• Full benefits including health, dental, vision, and 401(k) with 6% match
• Access to dedicated GPU compute resources for research and experimentation
• Daily catered breakfast and lunch
• Office in San Francisco, next to Oracle Park and the Caltrain