Explore my projects, the problems they address, how they work and the technologies behind them.
SynapseAI
Self-correcting multi-agent research system
A deployed GenAI research assistant that turns a topic into a structured, citation-backed report. A Researcher, a Critic and a Reviser agent iterate on every answer until it holds up.
A useful research report needs more than a fluent answer: it needs sources and a way to review the reasoning. SynapseAI brings research, criticism and revision into one workflow, turning a topic into a structured report with citations.
How it works
01
Research
The Researcher gathers material through live web search and Wikipedia retrieval to ground the report in external sources.
02
Review
The Critic reviews the answer as part of the three-agent loop, with attention to reasoning and citation quality.
03
Revise
The Reviser refines the answer through the feedback loop to produce a structured, citation-backed report.
NearFix
Hyperlocal services marketplace
A deployed marketplace for booking verified electricians, plumbers, tutors and more, with separate customer, provider and admin workflows plus transparent pricing.
Booking a local service involves finding a provider, understanding the price and coordinating a time. NearFix brings those steps together with separate workflows for customers, providers and administrators.
How it works
01
Find a service
Location detection and server-side geocoding support discovery of local services, including electricians, plumbers and tutors.
02
Book and schedule
Customers use the booking, scheduling and pricing flow to arrange a service.
03
Manage the work
Role-based dashboards separate customer, provider and administrator workflows. Clerk handles authentication, while the API covers bookings, providers, earnings and admin operations.
