I'm the kind of person who rarely stops at "it works." I usually end up asking how it works, why it works, what it's built on, and what happens if I break it.
I enjoy researching almost anything that catches my attention, dissecting systems to understand their internal architecture, and documenting what I learn along the way. I believe the best way to understand technology is to explore it deeply, question assumptions, and build things yourself.
I'm constantly looking for opportunities to learn, experiment, and understand the technology shaping the world around us.
- High-quality software engineering roles.
- AI/ML product development.
- Retrieval Augmented Generation (RAG) developments.
- Agentic AI development, Experimenting with Generative AI
- Open source collaboration and selective technical partnerships
Languages
AI / ML
NumPy · Pandas · Scikit-Learn
AI
Ollama · Agentic AI · RAG Pipelines · ChromaDB · FAISS · Pinecone · OpenAI · Gemini · Claude · Structured Prompting · Chain of Thought · Few Shot Learning
| Domain | Proficiency | Details |
|---|---|---|
| Supervised Learning | Inter | Feature engineering, model selection, evaluation, and deployment-aware validation. |
| LLM Systems | Inter | Prompt workflows, RAG, structured outputs, and product integration, LSTM, BLSTM |
| Data Processing | Inter | dataset curation, preprocessing |
| Agentic Systems | Inter | AI Agents, Pipelines and workflows, Automation |
Vortiq : Folder re-organizing 2.0
A lightning-fast, zero-cloud CLI utility that cleans your Windows directories by parsing file contents and visual data, automatically destroying clones and auto-generating intelligently named folder structures.
| Metric | Value |
|---|---|
| Stack | Python, Typer + Rich, QDrant DB, Scikit-Learn (DBSCAN Model) |
| Performance | Low-latency inference, offline first approach |
| Security | Comparison of hashes, HITL integration |
| Repository | github.com/44ompatil/vortiq |
IMDB Dataset Review Classifier
Analyze and predict the reviews for a movie, using Bidirectional LSTM
| Metric | Value |
|---|---|
| Stack | Python, Scikit-Learn, Keras (Bidirectional LSTM) |
| Performance | 90% Training Accuracy, 85% Validation Accuracy |
| Impact | Classify reviews using LSTM |
| Repository | github.com/44ompatil/review-classifier |
Web Scraping using LLM's
Get ready-to-go CSV sheets from any website using just a prompt.
| Metric | Value |
|---|---|
| Stack | Python, Ollama, Gemini, OpenRouter, Langchain, Crawl4AI |
| Scale | Hyperlink ingestion, schema generation, and text retrieval |
| Security | Dynamic Prompts |
| Impact | Scrape website data using local/cloud LLM |
| Repository | github.com/44ompatil/web_scraping |
June 2025 – Sep 2025
Contributed to the development of AI-powered educational solutions by building Retrieval-Augmented Generation (RAG) systems, designing data processing pipelines, and implementing vector-based knowledge retrieval architectures.
Focused on improving response quality, retrieval accuracy, and contextual relevance for domain-specific applications.
- Built an AI-powered chatbot using Retrieval-Augmented Generation (RAG) to provide college admission guidance and study material assistance through LLM-driven responses.
- Developed a data processing pipeline incorporating preprocessing, text cleaning, normalization, and context-aware chunking to enhance retrieval quality and downstream model performance.
- Generated semantic embeddings and implemented a vector database architecture using Qdrant for efficient indexing, storage, and similarity-based document retrieval.
- Applied prompt engineering and context injection techniques to improve response accuracy, relevance, and adherence to domain-specific requirements.
- Collaborated on the integration of LLM workflows and retrieval systems to deliver scalable and reliable AI-assisted user experiences.
Skills: #python #llm #rag #genai #langchain #qdrant #prompt-engineering #vector-databases #nlp #semantic-search
learning:
- CLI Tools
building:
- AI-enabled applications
- Tool for getting cutoff from college PDF's
exploring:
- retrieval-augmented generation
- Agentic AI and AI Agents
- ML Models
open_to:
- software engineering roles
- AI/ML work
- open source collaborations
- high-impact technical projects

