ML Engineer (Geometric Deep Learning & 3D Vision)

We are seeking a Machine Learning Engineer to solve complex spatial alignment and validation challenges. You will build the infrastructure to close the gap between CAD designs and real-world 3D reconstructions. The core of this role involves automating the "Ground Truth" process—developing sophisticated metrics to validate how digital components interface with organic (human bodies) or unstructured 3D environments.

Responsibilities

• 3D Registration & Alignment: Develop pipelines to align 3D meshes (photogrammetry) with CAD models using high-precision spatial transforms.
• Agentic Pipeline Orchestration: Build autonomous agents to manage the "whole flow"—from data ingestion and scale correction (mm vs. meters) to final metric validation.
• Data Integrity & Remediation: Architect automated systems to detect and correct common data pipeline failures, such as coordinate system mismatches, scale discrepancies (mm vs. meters), and metadata mislabeling.
• Closed-Loop Validation: Integrate alignment metrics directly into the ML inference flow, ensuring the model provides a confidence score or "alignment success" rating post-run.
• Spatial Feature Extraction: Extract actionable insights from the "whole flow" of provided data to optimize placement and interaction between objects.

Requirements

3D & Computer Vision

• Geometric Deep Learning: Proficiency with Open3D, PyTorch3D, or Trimesh for mesh processing and point cloud registration.
• Spatial Transforms: Deep understanding of Euclidean geometry, 3D coordinate systems, and photogrammetry workflows.

LLMs & Agentic Systems

• Agentic Frameworks: Experience building autonomous workflows using LangChain, LangGraph, AutoGPT, or CrewAI.
• Model Integration: Proficiency in prompt engineering and fine-tuning LLMs (OpenAI API, Anthropic, or local models via Ollama/vLLM) for structured data extraction and pipeline decision-making.
• Vector Databases: Experience with Pinecone, Milvus, or Weaviate for managing spatial embeddings and metadata.

Data Pipelines & DevOps

• Orchestration Tools: Expertise in building and monitoring pipelines using Dagster, Prefect, or Apache Airflow.
• Data Validation: Experience with Great Expectations or Pydantic to ensure data integrity across the "whole flow."
• Cloud Infrastructure: Familiarity with deploying ML workloads on AWS, GCP, or Azure using Docker and Kubernetes.
• Experience building "Human-in-the-loop" systems where LLMs handle the edge cases of 3D data processing.
• Background in Computational Geometry combined with modern LLM-Ops.
• A proven track record of automating complex, multi-step engineering workflows.
• Strong programming skills in Python is a must.
• Bachelor’s/Master’s degree in Computer Science/ Engineering or a related field.

We offer

• Opportunity to work on cutting-edge projects
• Work with a highly motivated and dedicated team
• Competitive salary
• Flexible schedule
• Benefits package - medical insurance, vision, dental, etc.
• Corporate social events
• Professional development opportunities
• Well-equipped office

About Us

Grid Dynamics (NASDAQ: GDYN) is a leading provider of technology consulting, platform and product engineering, AI, and advanced analytics services. Fusing technical vision with business acumen, we solve the most pressing technical challenges and enable positive business outcomes for enterprise companies undergoing business transformation. A key differentiator for Grid Dynamics is our 8 years of experience and leadership in enterprise AI, supported by profound expertise and ongoing investment in data, analytics, cloud & DevOps, application modernization and customer experience. Founded in 2006, Grid Dynamics is headquartered in Silicon Valley with offices across the Americas, Europe, and India.