Designing production-grade ML systems, scalable data pipelines, and applied AI solutions β from raw data to deployed models.
LinkedIn Google Scholar Instagram Email Profile Views
const abhinav = { role: "Senior Data Scientist & AI/ML Engineer", focus: ["Machine Learning", "Deep Learning", "Applied AI / GenAI", "MLOps"], currently: "Building scalable, production-grade ML systems & data pipelines", lifecycle: "framing β data pipelines β modeling β deployment β monitoring", toolbelt: "from GPU training clusters down to nginx configs & SQL warehouses", askMeAbout: ["RAG & Knowledge Graphs", "Agentic AI", "Forecasting", "Computer Vision"], funFact: "A well-tuned data pipeline is as satisfying as a clean model fit π" };
π§ Machine Learning & AI
- Supervised & Deep Learning
- NLP, RAG & Recommendation Systems
- Forecasting & Predictive Analytics
- Optimization & Decision Models
π€ Applied AI & GenAI
- LLM-powered assistants & integrations
- Agentic AI & workflow automation
- Knowledge-graph + vector retrieval
- Prompt & retrieval pipelines
ποΈ Computer Vision
- Image classification & segmentation
- Object detection & tracking
- Real-time inference pipelines
βοΈ Data Engineering & MLOps
- Large-scale ETL & feature pipelines
- Dimensional modeling & SQL tuning
- Containerized model serving
- Experiment tracking & monitoring
| Project | What it does | Stack |
|---|---|---|
| π€ RAG Knowledge Assistant | Retrieval-augmented Q&A over custom documents using embeddings + LLMs. | Python Β· LangChain Β· Qdrant |
| π Time-Series Forecasting | End-to-end forecasting with feature engineering, backtesting & model comparison. | Python Β· scikit-learn Β· XGBoost |
| ποΈ Real-Time Object Detection | Detection & tracking pipeline with a lightweight inference API. | PyTorch Β· OpenCV Β· FastAPI |
| π ML Model Serving API | Containerized inference service with experiment tracking & monitoring. | FastAPI Β· Docker Β· MLflow |
LinkedIn Google Scholar Instagram
Always learning, always shipping. β‘