Mahesh Solanki — twomathematicians-code
Engineering intelligent systems at the intersection of mathematics and code. From first principles to production — ML systems, GenAI, and domain solutions across BFSI, healthcare, and supply chain.
0
Repositories
0
Domains
0
Categories
0
Pipeline Stages
The title Unified Intelligence Engineer isn't marketing — it's a description of how I work. Most ML roles fragment intelligence into silos: data engineering here, model training there, deployment somewhere else.
I operate across the entire pipeline because the most impactful systems are designed holistically, not assembled from disconnected parts. Every project follows: Math → Data → ML → AI → Production → Impact.
Mathematical formulation comes first — Bayesian inference for credit risk, convex optimization for demand forecasting, structural equation modeling for causal analysis. Then the code follows. Never the reverse.
Every system derived from mathematical foundations — not copied from tutorials. Understanding why before how.
From hypothesis to production. Data pipelines, model training, API serving, monitoring — the full lifecycle, no handoff gaps.
Research is valuable only when shipped. Every project targets real users, real metrics, real impact — not just notebooks.
Patterns from BFSI inform healthcare models. Computer vision techniques enhance NLP pipelines. Intelligence compounds across domains.
Four interconnected pillars — each reinforcing the others.
Production ML
MLOps
Data Engineering
GenAI & RAG
NLP
Computer Vision
Statistics
Optimization
Linear Algebra
Custom-Built
Agents
Infrastructure
Meta-platform for AI system orchestration. Unified interface for managing multiple AI agents, workflows, and knowledge bases.
R-powered AI assistant bringing intelligent code generation and data analysis to the R ecosystem via natural language.
Research environment for deep computational experiments and mathematical exploration.
End-to-end production ML pipeline for supply chain demand prediction with monitoring and retraining.
Manifold-learning-based database system for high-dimensional data representation and querying.
Real-time fraud detection system for financial transactions using anomaly detection and classification models.
Production credit risk scoring pipeline with feature engineering, model training, and API deployment.
Comprehensive toolkit for causal inference methods — from propensity scores to instrumental variables.
Comprehensive natural language processing toolkit with pre-processing, modeling, and evaluation pipelines.
Production chatbot deployment framework with API serving, session management, and monitoring.
Multi-model sentiment analysis with benchmarking, fine-tuning, and production deployment options.
NLP research project exploring Chinese language processing and cross-lingual transfer learning.
Customer segmentation, churn prediction, and lifetime value analysis with ML-driven insights.
Collection of recommendation algorithms — collaborative, content-based, and hybrid approaches.
Analytics platform for exploratory data analysis with interactive visualizations and reporting.
Graph-based financial analysis system for detecting suspicious patterns in transaction networks.
ML-driven healthcare analytics for diagnostics, risk stratification, and operational optimization.
Time-series forecasting models for financial markets — stocks, FX, and macroeconomic indicators.
CV pipelines — object detection, face recognition, image segmentation using YOLO and OpenCV.
Variational Interpretable VAE research — making latent spaces interpretable for healthcare applications.
Research implementation of Bartlett correction for structural equation modeling fit statistics.
Curated collection of mathematical research implementations and explorations.
Framework for building and composing AI agents with tool-use, memory, and planning capabilities.
Swarm-intelligence-based crypto trading system with multi-agent coordination strategies.
Local LLM assistant integrated into RStudio via Ollama for private, in-IDE AI assistance.
Agent that processes and generates video content using NotebookLM-powered intelligence.
Comprehensive monograph on Conditionally Distributive Lattices in Markdown format.
Python implementations of advanced calculus concepts — limits, continuity, differentiation, integration.
Numerical methods implementations — root finding, interpolation, integration, ODEs in Python.
Numerical methods with C/C++ implementations for performance-critical computations.
Explorations in metric space topology — convergence, completeness, compactness with visualizations.
Hybrid recommender combining collaborative filtering, content-based, and deep learning for games.
Exploration of consciousness and cognition models — computational philosophy meets neuroscience.
Exploring computational paradigms beyond classical quantum computing models.
Neural network for eco-friendly driving pattern optimization and fuel efficiency prediction.
Likelihood ratio tests for structural equation modeling with advanced statistical implementations.
Bio-inspired optimization algorithm based on praying mantis hunting behavior.
Experimental project exploring novel computational approaches and algorithms.
NLP project applying computational methods to Sanskrit grammar (Vyakaraṇa) analysis.
Personal portfolio website — the one you're looking at right now.
Every system I build is rooted in mathematical reasoning. The research repositories aren't side projects — they're the ground truth that informs production decisions.
From Bartlett corrections in structural equation modeling to variational autoencoders for healthcare, from metric space topology to numerical methods in C++ — this is where the "unified" in Unified Intelligence begins.
Structural Equation Modeling
Bartlett corrections, likelihood ratio tests, fit statistics
Variational Inference
ViVAE — interpretable latent spaces for healthcare
Numerical Analysis
Python & C++ implementations of classical numerical methods
Abstract Algebra & Topology
CDL monograph, metric spaces, lattice theory
Advanced Calculus
Rigorous analysis with computational verification
Causal Inference
Propensity scores, IV methods, do-calculus
Whether you need a production ML system, an AI-powered product, or someone who thinks in math and ships in code.