I'm an AI Engineer and Researcher with a current focus on Generative AI, Reinforcement Learning and Computer Vision.
I'm most active on twitter, come say hi

I build AI systems that create impact through what I call "beautiful AI"

This beauty manifests in five key dimensions:

• bold ideas that resolve root problems
• thoughtful system architecture
• elegant and clear prompt pipelines
• data quality and validation-driven evaluation
• results that create impact

Do I consistently achieve this ideal balance? I must acknowledge that I don't - sometimes what appears as elegant design is actually concealing underlying disorder. Nevertheless, I'm passionate about finding clarity within complexity.

In past 6 years, I've worked across both AI research and engineering as a Researcher / ML Engineer / AI Engineer. I often find myself focusing on building systems that not only function efficiently but also deliver meaningful impact. I've learned to bridge research, practical knowledge, and creative thinking to instill real impact in products.

• Python, C++ (CUDA Programming), Node.js
• PyTorch, Multi-agent frameworks, DSPy, Hugging Face, OpenCV
• Multi-GPU Model Training, RL-driven Agents Training, LLM Training, Fine-tuning, Quantization, Abliteration
• Puhti/Mahti Server, Ray (Train, Tune, Serve), Skypilot, Model Deployment (Docker, vLLM, SGLang), A/B Testing, Observability
• Microservices, Autoscaling, LLM/CV System Design, Data Pipelines (retrieval, collection pipelines)
• AWS (S3, SQS, SageMaker, EC2, EKS, Lambda), GCP
• Django, FastAPI, Flask
• Architecting, designing, and writing prompt pipelines, prompt versioning, validation-driven evaluation pipelines