Senior AI Researcher, Agentic Systems
J
JobgetherConstruction AI
Remote work arrangement within the United States.Full-TimeSenior
SalaryBase salary: $184,000–$265,000 annually, depending on location, skills, experience, training, qualifications, and other business considerations.
Apply NowOpens the employer's application page
Job Details
- Experience
- At least 5 years of experience building machine learning systems
- Required Skills
- PythonMachine Learning
Requirements
- Master's or Ph.D. in a relevant field, or equivalent professional experience.
- At least 5 years of experience building machine learning systems.
- Experience taking at least one LLM- or VLM-based system into production and maintaining it over time.
- Ability to design evaluation frameworks from scratch when reliable ground truth is sparse, costly, or contested.
- Strong knowledge of multimodal AI, including vision-language models, embeddings, retrieval systems, context engineering, and fine-tuning.
- Strong software engineering skills, particularly in Python, asynchronous orchestration, tracing, and cost instrumentation.
- Ability to build custom testing and evaluation harnesses for AI systems.
- Familiarity with spatial data concepts, including coordinate frames, camera poses, and projections.
- Strong technical judgment about automation suitability and system reliability.
- Ability to collaborate with research and engineering teams, communicate findings, and translate experimental results into production-ready solutions.
- Preferred: experience with post-training techniques or reinforcement learning applied to agent trajectories.
- Preferred: familiarity with document understanding for technical drawings, specifications, and structured PDF documents.
Responsibilities
- Design and develop agents that plan and execute multi-step tasks across imagery, spatial geometry, and technical documents.
- Build and maintain the context and retrieval layer for spatially and temporally indexed construction records.
- Develop tools and interfaces for locating components, comparing site captures, inspecting drawing regions, and querying BIM elements.
- Design evaluation frameworks for complex tasks with sparse, costly, or subjective ground truth.
- Use evaluation results to assess system readiness, identify limitations, and guide improvements before production deployment.
- Select models for each task, balancing cost, latency, and accuracy across frontier-model APIs, open-weight models, fine-tuned models, and computer vision techniques.
- Connect generated claims to supporting evidence in images, drawing sheets, or specification sections.
- Prototype approaches, discontinue ineffective solutions, and collaborate with platform engineers to productionize successful systems.
- Monitor and optimize deployed systems for reliability, performance, and cost efficiency.
View Full Description & ApplyYou'll be redirected to the employer's site
