Senior Scientist on the AWS FDE team. I publish applied AI research and embed with engineering teams to build custom AI systems, from scoping and experimentation to fine-tuning and production deployment.
8+ years in applied AI/ML · Focused on LLMs and agentic systems
Latest BlogBuilder Center
Language models solved only 34% of SWE-InfraBench's infrastructure-as-code tasks in one attempt. We ran Kiro over all 100 and found that a full agent loop with Sonnet 5 reaches 82%, with partially correct answers converging once the agent can re-run the tests itself.
View all 29 blog posts
Latest TalkGITEX AI · Berlin
Live demo of the IDP Accelerator, a scalable, serverless solution for automated document processing and information extraction on AWS. The asset combines generative AI and optical character recognition (OCR), using services such as Amazon Bedrock Data Automation and Amazon Bedrock foundation models to extract, classify, and process documents at scale.
View all 35 talks
Latest PaperICML 2026 Workshop on Weight-Space Symmetries
Multi-task model merging combines separately trained expert models into a single model that handles all tasks without co-training. Standard practice merges experts at their optimal validation loss. We challenge this convention by systematically studying how training duration of domain experts affects the quality of the merged model. We fine-tune experts on five domains (Math, Code, Instruction Following, Multilingual, and Safety) across three model sizes (Qwen 3.5 0.8B, 2B, and 4B), saving checkpoints from 25% to 500% of the optimal training steps and evaluating five merging methods at each duration. Our findings reveal a striking method-dependent pattern: simple averaging degrades sharply with overfitting, while sparsification-based methods achieve their best performance well past the validation optimum. We formalize this through bias-variance decomposition analysis, drawing a parallel to random forests where averaging benefits from high-variance individual learners. These results suggest that training duration and merging method should be chosen jointly rather than independently.
View all 17 papersAgentic code scanning system that found 30+ high-priority defects in a production automotive codebase.
Agentic pipelines that recovered the business logic buried in 1.3 million lines of mainframe code, now moving to production rollout.
Test-time self-reflection for content localization across 17 European markets, tuned for quality, cost and latency.
Open to collaboration, speaking invitations, and conversations about applied AI.