About the Role
We are seeking a Principal Machine Learning Engineer – Model Efficiency & Optimization to serve as the technical anchor for ABBYY’s model optimization strategy.
This is a senior individual contributor role for a deep domain expert who will define how ABBYY builds efficient, high-performing, production-ready models for document AI at scale. You will set technical direction from research exploration through production deployment, combining strong theoretical expertise with hands-on implementation.
Key Responsibilities
Research Direction & Technical Strategy
- Own the end-to-end technical direction for model efficiency and optimization, from research agenda to production deployment
- Define approaches for building efficient, production-ready models optimized for document AI use cases
- Establish frameworks for evaluating quality vs. efficiency trade-offs (accuracy, latency, memory footprint)
- Set standards for what constitutes a successful optimized model across document understanding benchmarks
- Evaluate and adopt emerging techniques in model optimization and compression
- Influence modeling strategy across teams by integrating efficiency-first thinking into model development
Hands-on Implementation & Experimentation
- Lead design and implementation of optimization pipelines, training objectives, and compression techniques
- Run large-scale experiments and analyze training dynamics, instabilities, and capability gaps
- Develop novel optimization approaches tailored to document understanding tasks, including layout and multimodal challenges
- Diagnose and resolve failure modes such as quality degradation and generalization gaps
- Prototype and validate new techniques before scaling to production training runs
Cross-Functional Collaboration
- Partner with the Document AI Data team to define training data requirements for optimized models
- Collaborate with Platform teams on distributed training infrastructure, experiment tracking, and compute strategy
- Work closely with Modeling teams to ensure optimized models meet quality and performance standards
- Communicate technical trade-offs, findings, and recommendations clearly to engineering, product, and leadership stakeholders
Qualifications
Education & Experience
- MS or PhD in Computer Science, Engineering, Mathematics, or related field (PhD preferred)
- 10+ years of experience in Machine Learning / AI, with focus on:
- Model optimization
- Efficient deep learning
- Large-scale model deployment
- Demonstrated track record of contributions to model efficiency (e.g., publications, patents, or industry impact)
- Proven experience optimizing large-scale language and/or vision models for production
- Deep understanding of trade-offs between model quality, size, and inference performance
Technical Expertise
- Deep expertise in model optimization and compression techniques (e.g., quantization, pruning)
- Strong knowledge of efficient deep learning methodologies
- Expertise in Vision-Language Models (VLMs) and multimodal optimization challenges
- Strong programming skills in Python and deep proficiency with PyTorch or similar frameworks
- Experience with distributed training systems and large-scale experimentation workflows
- Strong evaluation methodology for optimized models, including benchmarking, efficiency profiling, and regression analysis
Leadership & Communication
- Recognized technical authority in model optimization or efficient AI systems
- Proven ability to influence technical direction without formal authority
- Strong track record of driving applied research → production impact
- Excellent communication skills, with the ability to clearly articulate complex trade-offs
- Collaborative mindset with the ability to align cross-functional teams