Sachin Shelar

I hold a postgraduate degree in Machine Learning with additional specializations in Deep Learning, NLP, and Agentic AI. Over the past several years, I have led multiple end-to-end AI solution initiatives—from research and prototyping to deployment and scaling. My work includes replacing legacy OCR systems like ABBYY FlexiCapture with advanced transformer-based models such as LayoutLM, Donut, and DETR for document and signature recognition. These projects have significantly improved accuracy, reduced manual intervention, and lowered operational costs.

I am a hands-on Python engineer and have built numerous AI-driven applications using frameworks such as Flask, Django, fastAPI and Streamlit, later converting them into robust production services through CI/CD automation, comprehensive testing, and well-structured API layers (Swagger).

In my current role, I have also worked with Grafana for observability—creating dashboards for model performance, API latency, and service uptime. Alongside this, I have integrated logging-as-a-service, enabling centralized log collection, alerting, and real-time monitoring across distributed components. These capabilities have helped ensure reliability, traceability, and rapid debugging in production environments.

My recent work includes benchmarking Azure Document Intelligence against our existing LLM based solutions, further demonstrating my commitment to delivering intelligent, scalable, and business-aligned AI systems.

I take pride in my collaborative approach—partnering with DevOps teams, data engineers, and business stakeholders to ensure seamless deployment and measurable business outcomes. I am eager to bring this blend of technical depth, innovation, and practical problem-solving to your organization.

Demo projects on my personal youtube channel

(use of livekit with my agentic application)

(voice activated Agentic RAG)