Building intelligent systems from research to production.
Designing and building end-to-end AI solutions including Agentic RAG systems, multimodal document understanding, computer vision models, and data-driven IoT analytics. Ph.D. in Computer Science with a strong engineering foundation.
Designing and implementing end-to-end RAG systems and production AI solutions. Working with LLMs, OCR, NER, multimodal models, and vector databases. Building scalable FastAPI services and resilient data extraction pipelines for production deployment.
Developing architectures for managing, decoding, and translating data from WMBus, NB-IoT, and MQTT protocols, embedding AI modules for anomaly detection and predictive analytics on time-series. Designing and developing Android applications for WMBus data collection and workforce management, integrating AI modules for image analysis and OCR-based text recognition.
Designing and developing unsupervised domain adaptation algorithms for object detection and deep learning algorithms for processing egocentric data captured through Mixed Reality devices.
Designing and developing an anomaly detection system integrating forecasting and anomaly detection models through a voting-based decision mechanism on semiconductor manufacturing time-series.
Python, PyTorch, PyTorch Lightning, TensorFlow, Hugging Face, Scikit-learn, OpenCV, LangChain, LangGraph
FastAPI, PostgreSQL, MongoDB, Redis, Pandas
Docker, Git, Linux, AWS, CI/CD, MLflow, Jupyter
Java, C++, C#, JavaScript, TypeScript, Flutter
A modular Agentic RAG built with LangGraph — learn and implement Retrieval-Augmented Generation Agents in minutes with multi-agent workflows.
Open-source toolkit for reliable RAG pipelines: convert PDFs to Markdown, clean documents, inspect chunks, compare chunking strategies, and enrich metadata for LLMs.
Detectron2 implementation of Domain Adaptive Faster R-CNN for Object Detection in the Wild, addressing domain shifts in computer vision (CVPR).