Sk Sohel

I’m an AI/ML Engineer with production experience at Infosys Springboard, where I architected and deployed a real-time YOLOv8 vision pipeline. It delivered 89.6% mAP and reduced manual audits by 70%, while I kept the system performance-focused and monitored.

I specialize in computer vision and LLM-powered systems—especially multi-agent architectures and intelligent data platforms. I’ve built complete inference pipelines (OpenCV → Flask API → Streamlit dashboard) with production-grade monitoring and alerts.

In my projects, I designed a self-healing multi-agent pipeline using LangGraph with a stateful critique-revision loop, parallel LLM runs, Redis caching with TTL, exponential backoff, and fallback chains. I also delivered a full-stack AI data explorer that integrates Groq for natural-language-to-SQL, automated charts, anomaly detection, and executive narrative summaries.

I enjoy turning messy real-world data into reliable products that are fast, resilient, and privacy-aware. I’m now seeking MLE / Applied AI roles where I can keep building impactful, deployable AI systems.