I build intelligent systems that are useful, efficient, and easy to understand.
Exploring the space between applied AI, machine learning research, and thoughtful software engineering.
I’m an AI/ML builder with a strong interest in turning ideas into working products and reproducible experiments. My projects range from tool-using LLM applications and multilingual NLP to computer vision, scientific imaging, and small models designed to run on everyday hardware.
I enjoy the full journey: understanding the problem, designing the data and model, making inference practical, validating the result, and documenting the work so someone else can learn from it.
|
Agentic applications, tool use, retrieval-friendly services, and human-centered interfaces. |
Compact models, ONNX export, INT8 quantization, and offline inference for edge devices. |
Deep learning, scientific imaging, visual recognition, and experiments that make complex ideas clearer. |
| Project | What it demonstrates | Stack |
|---|---|---|
| Agentic MCP Chatbot | An MCP-powered chatbot where an LLM discovers and calls tools for card search, eligibility, rewards, and spending questions. | Python · FastAPI · FastMCP · OpenAI · Docker |
| Offline Partner Assistant | Small, fully offline models for Hinglish command parsing, smart replies, and address parsing, exported to ONNX and INT8-quantized. | PyTorch · ONNX · NLP · Edge AI |
| DeepTalk.ai | An interactive chatbot designed to make deep learning concepts easier to explore and learn. | Python · PyTorch |
| VideoGist | A multilingual YouTube summarizer with transcription, recipe extraction, and an interactive Streamlit interface. | Python · Streamlit · OpenAI · Groq |
| Food Vision Classification | A transfer-learning image classifier trained to recognize 101 food categories from the Food101 dataset. | TensorFlow/Keras · CNNs · Jupyter |
Browse all public projects and repositories
This archive includes project work, experiments, notebooks, and learning repositories.
| Languages | Python |
| AI & ML | PyTorch · TensorFlow/Keras · OpenAI API · Groq · computer vision · NLP · transfer learning · model compression |
| LLM systems | MCP · FastMCP · tool-calling agents · FastAPI · session-aware conversations |
| Data & scientific computing | NumPy · pandas · SciPy · scikit-image · Zarr · Matplotlib · Jupyter |
| Deployment & workflow | ONNX · INT8 quantization · Docker · uv · pytest · Kaggle · Git/GitHub |
- Making AI models run efficiently on any device.
- AI hardware and hardware–software co-design for efficient intelligent systems.
If you’re working on applied AI, machine learning systems, efficient inference, or research-oriented software, I’d be happy to connect and learn from your work.
Thanks for stopping by — feel free to explore, learn, and build alongside me.