Projects
A working portfolio of 50+ AI and machine learning projects, from securing autonomous agents to shipping clinical and financial systems. Filter by domain to see depth across healthcare, finance, cybersecurity, agentic AI, and the enterprise.
Securing agents, non-human identities, and the AI tool layer for enterprises.
Autonomous and multi-agent systems that plan, use tools, and get real work done.
Clinical assistants, diagnostic models, and patient-facing AI built for a regulated space.
Financial RAG, fraud detection, and risk modeling for banking and markets.
Knowledge graphs, semantic search, and LLM platforms wired into real business data.
Generative pipelines for writing, video, audio, and research at scale.
LLM-powered tooling that speeds up how people build, apply, and ship.
Forecasting, computer vision, and full-stack products shipped to real users.
Non-human identity governance for the agent era
SaaS platform that discovers, risk-scores, and auto-rotates non-human identities (service accounts, API keys, agent credentials) across AWS, Azure, GCP, and GitHub. Closes the fastest-growing gap in enterprise AI security: the exploding population of machine identities that agents create and use.
The 'npm audit' for Model Context Protocol servers
A security scanner for MCP servers that detects tool poisoning, rug-pull tool definitions, and prompt-injection payloads hidden in tool metadata before an agent ever trusts them. Purpose-built for the MCP supply-chain attack surface that is now the most urgent problem in agentic AI security.
AI-powered external attack surface management
External attack surface management and cyber threat intelligence platform that maps an organization's internet-facing footprint and enriches it with AI-driven risk analysis, giving security teams a live, prioritized view of exposure.
AI assistant for security compliance and audit
LLM assistant that accelerates security reviews and audit-readiness by mapping controls, drafting evidence, and answering compliance questions against frameworks, cutting the manual effort in SOC 2, ISO, and enterprise vendor assessments.
AI misinformation and fake-content detector
Detection system that flags AI-generated and manipulated content, helping teams defend against misinformation and synthetic-media fraud, an increasingly weaponized threat vector for enterprises.
AI-assisted originality and plagiarism checker
Semantic plagiarism detection that goes beyond string matching to catch paraphrased and AI-rewritten copying using embedding-based similarity.
Multi-agent system that writes winning RFP responses
Agentic RFP autopilot that decomposes a request for proposal, researches answers across a knowledge base, and drafts compliant, on-brand responses with a coordinated team of specialized agents, collapsing days of proposal work into hours.
Stateful multi-agent orchestration with LangGraph
A multi-agent orchestration framework built on LangGraph, wiring specialized agents into a stateful graph with shared memory, tool routing, and human-in-the-loop checkpoints, a reference architecture for production agent workflows.
Collaborative agent workflows on LlamaIndex
Multi-agent workflows built with LlamaIndex, coordinating retrieval, reasoning, and tool-using agents over enterprise documents to answer complex, multi-step questions.
Agents that collaborate to automate tasks end to end
A multi-agent system for collaborative task automation, where planner, worker, and reviewer agents divide a goal, execute in parallel, and reconcile results.
Autonomous agent that reviews pull requests
An autonomous code-review agent that reads diffs, reasons about correctness and security, and posts actionable review comments, bringing consistent senior-level review to every pull request.
Autonomous agent that hunts and ranks jobs
An autonomous agent that continuously scouts job boards, matches roles against a candidate profile, and ranks the best-fit opportunities without manual searching.
AI-first healthcare experience
A healthcare application that brings AI into the patient journey, from intake and triage to guidance, designed for the trust and clarity a clinical setting demands.
Grounded clinical Q&A over medical knowledge
A retrieval-augmented chatbot that answers patient-care questions grounded in a curated medical knowledge base, reducing hallucination risk in a domain where accuracy is non-negotiable.
CNN skin-cancer classifier from dermoscopic images
A custom convolutional neural network that classifies melanoma and other skin conditions from dermoscopic images, applying deep learning to early cancer screening.
MCP server for a wellness platform
A Model Context Protocol server for the BreatheIT wellness platform, exposing platform data and actions to AI agents through the emerging standard for tool integration.
Retrieval-augmented AI for financial documents
A financial RAG system that answers questions over filings, reports, and market documents with source-grounded citations, built for the accuracy and traceability finance workflows require.
ML fraud detection on imbalanced transaction data
A machine-learning pipeline that detects fraudulent banking transactions on highly imbalanced data, using resampling and ensemble models to maximize recall without drowning analysts in false positives.
AI-assisted equity analysis
A tool that analyzes stocks by combining market data with AI-generated summaries and signals, turning raw financials into readable insight.
Credit-risk EDA on consumer loans
An exploratory data analysis of Lending Club loan data identifying the driving factors behind loan default, the kind of credit-risk analysis that underpins lending decisions.
Regression modeling for property valuation
A regression model that predicts housing prices from property and market features, covering the full workflow from feature engineering to model evaluation.
Knowledge graphs built over organizational data
A builder that constructs an enterprise knowledge graph over an organization's data, connecting people, documents, and entities so AI systems can reason over relationships instead of isolated chunks.
Semantic search that understands intent
An AI-powered semantic search application that retrieves by meaning rather than keywords, delivering relevant results across large document and product corpora.
Enterprise chatbot on Azure OpenAI
A production-oriented conversational assistant built on Azure OpenAI, demonstrating enterprise deployment patterns: managed keys, grounding, and controlled responses.
LLM-powered conversational assistant
A flexible LLM chat assistant scaffold with streaming responses and tool hooks, a clean foundation for domain-specific copilots.
Shopping assistant for online storefronts
A conversational commerce assistant that helps shoppers find products, answer questions, and complete purchases, lifting conversion through natural-language discovery.
Smart chunking for LLM pipelines
A document and context splitter that chunks source material intelligently for LLM pipelines, preserving structure and semantics to improve retrieval quality.
AI-driven SEO analysis and optimization
An AI SEO tool that audits pages, surfaces optimization opportunities, and generates recommendations to improve search visibility.
AI-assisted authoring and delivery
A newsletter platform that uses AI to draft, format, and personalize issues, then handles delivery, an end-to-end publishing workflow.
AI long-form content writer
An AI article and content writer that researches a topic and produces structured, on-brand long-form drafts, built to scale content production without losing quality.
Turn news articles into social posts
A pipeline that ingests news articles and generates platform-tailored social-media posts, automating the news-to-distribution loop.
AI video creation pipeline
An AI video creation pipeline that assembles scripts, voiceover, and visuals into finished videos, compressing production into an automated workflow.
AI video translation and dubbing
A tool that translates and re-voices video into other languages, combining transcription, translation, and speech synthesis to localize content automatically.
AI transcription and note-taking
An AI transcription tool that turns speech and audio into accurate, structured text and notes, built for meetings, interviews, and research.
AI assistant for reading research papers
An assistant that reads, summarizes, and answers questions about academic papers, helping researchers digest dense literature faster.
Interactive prompt-engineering workbench
An interactive builder for crafting, testing, and refining prompts, giving teams a structured way to engineer and version their prompts.
AI that explains unfamiliar code
A tool that reads source code and produces clear, human explanations of what it does and why, accelerating onboarding and code comprehension.
Turn unstructured text into structured data
An LLM-powered converter that extracts structured CSV from messy, unstructured text, a practical bridge from documents to databases.
AI brainstorming partner
An AI idea-generation tool that expands a seed prompt into a diverse set of concepts, a fast brainstorming partner for product and content teams.
Browser extension for instant page Q&A
A browser extension that summarizes any web page and answers questions about it in place, bringing an LLM copilot to the reading experience.
Intelligent job-application autofill
A browser extension that intelligently autofills job applications from a candidate profile, eliminating the repetitive data entry of the job hunt.
AI job-search and application assistant
An AI assistant that streamlines the whole job search, matching roles, tailoring materials, and tracking applications in one place.
AI sales-prospecting assistant
An AI sales-prospecting assistant that researches leads and drafts personalized outreach, helping sales teams find and engage the right accounts faster.
Forecasting repository activity with LSTM & Prophet
A time-series forecasting system that predicts repository activity and trends using TensorFlow/Keras LSTM networks alongside Facebook Prophet and StatsModels, benchmarking classical and deep-learning forecasters.
Deep-learning OCR for handwriting
A deep-learning model that recognizes handwritten text from images, applying computer vision and sequence modeling to the classic OCR challenge.
Graph analytics on social and well-being data
An online social-network analysis project that studies world-happiness data through graph analytics and community detection, extracting insight from network structure.
Live app for discovering San Francisco events
A deployed web app that surfaces and recommends San Francisco events, pairing a clean product experience with AI-assisted discovery.
Live job-discovery web app
A deployed job-discovery application that helps users search and filter opportunities through a fast, modern interface.
Restaurant discovery app
A restaurant-discovery app that helps users find places to eat with smart filtering and a clean, responsive interface.
AI movie recommendation app
A movie-discovery app that recommends films based on user preferences, combining a content catalog with AI-driven suggestions.
I lead AI engineering and security for enterprises and startups, from first prototype to production. Let's talk about what you are building.