I bridge the gap between Data Science innovation and Production reliability. With over 10 years of experience (3.5+ in MLOps), I specialize in building scalable, secure, and self-healing AI platforms for enterprise retail clients.
- 🔭 I’m currently working on MLOps & Edge AI Governance
- 🌱 I’m currently learning Advanced RAG Architectures & Rust for MLOps
- 💬 Ask me about Azure ML, Databricks, Model Governance, and Cost Optimization
- 📫 How to reach me: sunilkunchoor@gmail.com
| Project | Description | Tech Stack |
|---|---|---|
| 🚦 MLOps Traffic Light | Automated Model Governance System A GitHub Action that acts as a "Gatekeeper" for PRs. It validates Code Quality, Security (Snyk/Semgrep), and Model Performance before allowing a merge. |
Python GitHub Actions PyTest Snyk |
| 🧞 AdGenie LLMOps | GenAI Pipeline with "Prompts as Code" An end-to-end LLM lifecycle management system. Features automated evaluation loops using GPT-4 as a judge to grade prompt changes. |
LangChain MLflow OpenAI Azure |
| 👁️ Retail-Lens | AI-Powered Smart Shelf Assistant Empowers employees to quickly identify and rectify issues like out-of-stock items, misplaced products, and incorrect price tags, ultimately improving store efficiency and customer satisfaction. |
Azure Vision Docker OpenCV |
Cloud & Infrastructure
MLOps & Data
CI/CD & Governance
| Credential | Issuer | Badges/Portfolio |
|---|---|---|
| 🏅 **Databricks Data, ML, and GenAI ** | Databricks | Link |
| 🏅 **IBM Data Science Specialization ** | Coursera | Link |
| 🏅 GitHub Actions | GitHub | Link |
| 🎓 Post Graduate Program in AI & ML | University of Texas, Austin (Great Learning) | Link |
| 🎓 Master of Science (Mathematics) | Bangalore University | - |
I believe that MLOps is about Engineering Rigor, not just writing scripts. My work is guided by four principles:
- Zero-Friction for Data Scientists: My goal is to abstract away the infrastructure (Kubernetes, Docker) so Data Scientists can focus purely on Modeling and Mathematics, not plumbing.
- Structure Begets Speed: Ad-hoc scripts don't scale. I enforce strict project structures and CI/CD templates to ensure that every deployment is repeatable, audit-ready, and automated.
- Guardrails Enable Confidence: Strict governance (like "Traffic Light" validation) allows teams to deploy faster, knowing that safety checks are baked into the pipeline.
- Frugal Architecture: Cloud costs should not grow linearly with model usage. I prioritize optimized inference (ONNX/Quantization) to keep bills low.