About
Technical judgment for ML systems that need to work.
I am a machine learning scientist and engineer working across high-performance ML, NLP, evaluation, and production systems.
My work moves between research, implementation, and advisory review: optimizing large-scale training and inference, evaluating model behavior, and helping teams reason clearly about reliability, performance, model behavior, and deployment risk.
fairgradient.ai is the formal home for that work: technical advisory, public writing, speaking, and selective mentorship for engineers building serious AI systems.
Technical practice
What I work on
The foundation behind the advisory practice: applied ML systems, evaluation, performance, and the engineering discipline needed after a model leaves a notebook.
ML System Optimization
Deep experience with quantization, distillation, and kernel-level tuning to push inference and training performance at scale.
Evaluation & Responsible Deployment
Hands-on work in model evaluation, bias detection, red-teaming, and deployment practices for real-world systems.
Model Architecture & Fine-Tuning
From dataset curation to PEFT/LoRA fine-tuning, building specialized models that solve problems off-the-shelf solutions can't.
MLOps & Production Systems
Building evaluation harnesses, CI/CD pipelines, and serving infrastructure that keep ML systems reliable after launch.
Stack
Technical stack
Core Expertise
- LLM Fine-tuning
- Distributed Training
- Algorithmic Fairness
- Inference Optimization
- Causal Inference
Engineering Stack
- PyTorch
- JAX
- CUDA
- TensorRT
- Kubernetes
- Ray
- vLLM
Cloud & Ops
- AWS SageMaker
- GCP Vertex AI
- MLflow
- Weights & Biases
- Terraform
Recognition
Recognition
Globee Gold Award: AI Professional of the Year
Highest score in category, recognizing contributions to responsible AI, large-scale ML systems, and industry thought leadership.
Globee Silver Award: Most Innovative AI Person of the Year
Recognized for innovation in AI fairness, inference optimization, and community impact through mentorship and publications.