Building systems that bridge mathematics and machine learning
Open to opportunities in ML research, theorem proving, and symbolic AI. Let's chat →
I'm a researcher working at the intersection of mathematics and machine learning. Currently at Internet Brands, I build systems that make sense of complex data—whether that's filtering attorney databases, scraping, analytics and other half of me is focused on developing, reading about symbolic AI techniques for theorem proving and mathematical reasoning.
My background is in pure mathematics (M.Sc. from University of Mumbai), and I'm particularly interested in automated theorem-proving systems like Lean. These days, I spend my time tinkering, reading Lean4 etc. about how to make AI systems more reliable through mathematical rigor. And develop systems that combine neural and symbolic approaches.
May 2025 – Present
Building a neuro-symbolic math solver that combines LLM response with SymPy for symbolic validation. The goal is to create reliable mathematical reasoning by bridging neural and symbolic approaches. Future roadmap includes integrating theorem provers like Lean or Coq for formal verification of results. As I believe with some testing of Harmonic Aristotle Lean APIs this can be done.
A contextual chatbot using Langchain, Pinecone Vector Database, and GPT-3.5. Implemented multi-lingual embeddings for semantic search across languages. Built to explore how vector databases handle cross-lingual retrieval.
Built ML models and OCR tools for automating real estate transaction analysis. Developed web crawling systems for multi-source data validation. Reduced manual data entry by 40% through computer vision techniques.
Architected data pipelines processing millions of legal records. Focused on classification and evaluation metrics for structured legal data. Improved team efficiency by 30% through workflow optimization.
Category Theory by Tom Leinster
Exploring the abstract structures that connect different areas of mathematics—relevant for understanding compositional approaches in AI systems.
I studied mathematics at University of Mumbai, focusing on functional analysis, abstract algebra, and optimization. During my master's, I was a summer research fellow at IISER Thiruvananthapuram working on metric space completion theory.
After graduating, I taught graduate-level mathematics before transitioning into ML research. The mathematical training has been invaluable—particularly when working on symbolic reasoning systems or optimizing large-scale pipelines.
Certifications: Machine Learning (Stanford/Coursera - Andrew Ng), Advanced SQL & EDA with Python (DataCamp)
Languages: Python, SQL, PyTorch, PySpark, Octave, LaTeX, C
ML/AI: PyTorch, scikit-learn, Langchain, HuggingFace, Vector DBs (Pinecone), Streamlit
Engineering: Git, GCP, Postman, Selenium, PostgreSQL, Tableau
I'm always interested in conversations about mathematical reasoning, symbolic AI, or theorem proving. Feel free to reach out via email or check out my blog where I write about math and ML.
Available for full-time positions and research collaborations