Myself Arjun Jagdale, turning research into production-ready ML systems. I'm an AI engineer who codes at the intersection of deep learning research and production engineering — building everything from anti-spoofing CNNs to parameter-efficient transformers, while actively contributing to core Hugging Face libraries.
class ArjunJagdale:
def __init__(self):
self.role = "AI Engineer & Open Source Contributor"
self.focus_areas = ["Deep Learning", "Computer Vision", "NLP", "MLOps"]
self.current_work = "Contributing to Hugging Face core libraries"
self.interests = ["RAG Systems", "Model Compression", "Cloud-Native ML"]
def get_expertise(self):
return {
"frameworks": ["PyTorch", "HuggingFace", "scikit-learn", "TensorFlow"],
"specializations": ["Parameter-Efficient Fine-Tuning", "CNNs", "Transformers"],
"cloud": ["IBM Cloud", "Google Cloud", "Docker", "Kubernetes"],
"tools": ["LangChain", "LlamaIndex", "Gradio", "Git"]
}
Resolved compatibility issue with NumPy 2.0+ by wrapping stratify column array access with np.asarray(). Maintains backward compatibility with NumPy 1.x while fixing array copy errors.
Updated docstrings for add_column(), select_columns(), select(), filter(), shard(), and flatten() to clarify that these methods return new datasets instead of modifying in-place. Significantly improves API documentation clarity.
Added validation check in FolderBasedBuilder to prevent silent fallback to current directory when loading folder-based datasets without required parameters. Improves user experience by catching errors early.
Implemented support for Date, UTCDate, and UTCTime features in Croissant schema generation. Automatically infers correct dataType (sc:Date, sc:Time, or sc:DateTime) based on format string.
Eliminated shared mutable default values in dataclass fields by replacing helper functions with explicit constant copies. Makes configuration behavior more explicit and prevents subtle bugs.
Implemented comprehensive unit tests for cache retrieval function covering successful cache hits, missing cache scenarios, and error status handling. Improves code coverage and reliability.
Removed redundant custom implementations of update_repo_settings() across test utilities by leveraging official huggingface_hub API. Cleaned up 222 lines of code while maintaining full functionality.
Custom CNN Architecture for Real vs. Spoof Detection
results = {
"validation_accuracy": "99.39%",
"test_accuracy": "99.29%",
"dataset_size": "38K images",
"parameters": "4.88M"
}
Tech: PyTorch • OpenCV • MediaPipe • Gradio
Low Rank Adaptation Fine-Tuning for Named Entity Recognition
efficiency = {
"parameter_reduction": "99.49%",
"trainable_params": "641,347",
"f1": "0.6744",
"Precision" : "0.6539"
"speedup": "750x faster training"
}
Tech: PyTorch • HuggingFace • LoRA • NER
Semantic Retrieval & QA Over YouTube Comment Corpora
pipeline = {
"embedding_model": "all-MiniLM-L6-v2",
"index": "FAISS",
"retriever_k": 10,
"splitter": "RecursiveCharacterTextSplitter"
}
Tech: LangChain • FAISS • HuggingFaceEmbeddings • RAG • OpenRouter
Graph-Orchestrated Retrieval & QA Across Arbitrary Web Sources
workflow = {
"orchestrator": "LangGraph StateGraph",
"memory": "MemorySaver",
"vector_store": "FAISS",
"ui": "Gradio (configurable lengths)"
}
Tech: LangChain • LangGraph • FAISS • HuggingFaceEmbeddings • Gradio • RAG
Bidirectional Sequence Modeling for IMDB Reviews
performance = {
"test_accuracy": "85.19%",
"train_test_gap": "4%",
"architecture": "2-layer BiLSTM",
"embedding_dim": 128
}
Tech: PyTorch • LSTM • Gradio • NLP
Classical ML with Advanced Feature Engineering
model_comparison = {
"Random_Forest": "97.2%",
"Logistic_Regression": "96.5%",
"SVM": "95.8%",
"after_tuning": "+1-2%"
}
Tech: scikit-learn • Pandas • Matplotlib • NumPy
Published research on RAG systems that dynamically learn from web content, combining retrieval mechanisms with adaptive learning strategies for improved information access and knowledge synthesis.