Smriti / स्मृति
A personal memory app that helps people capture everyday details and retrieve them later using natural language — including voice and photo-based memories.
I'm Jhanvi — a backend engineer with 9+ years of experience across fintech, streaming media and enterprise systems. I build Java microservices, event-driven platforms and reliable integrations, and I enjoy taking independent AI products from an idea to a working application.
Independent products are where I explore the full engineering lifecycle — from shaping the problem to architecture, implementation and iteration.
A personal memory app that helps people capture everyday details and retrieve them later using natural language — including voice and photo-based memories.
An AI-assisted job discovery and matching workflow designed to find relevant roles, remove duplicates and explain how closely each opportunity fits a candidate profile.
A few engineering problems from my work across product and enterprise environments — focused on system behaviour, trade-offs and outcomes.
Worked on service-to-service data synchronisation using Debezium change data capture and Kafka. The event-driven approach reduced reliance on an intermediary message-processing hop and helped simplify the path for propagating subscription entitlement changes.
Enhanced an application that displays customer invoices, integrating ERP events and invoice documents while separating ingestion concerns from the display experience. Improved API response time by 20% and strengthened access control around protected operations.
Built and enhanced reporting services, REST APIs and asynchronous integrations. Worked across report generation, archival, message consumption and workflow orchestration in a banking environment.
Built and maintained Java services in large enterprise environments, developing a foundation in API design, batch processing, integration patterns, production support and collaborative delivery.
The diagram is intentionally conceptual. Add the exact topic, schema, retry and failure-handling details you can comfortably explain in an interview.
Grouped by the kind of engineering problem each technology helps solve — not a wall of logos.
Short technical write-ups drawn from practical engineering problems. These can grow into deeper articles over time.
Why similarity scores alone can return plausible but irrelevant memories — and how metadata, tag overlap and state filters can improve retrieval precision.
Exploring the trade-offs of capturing database changes from a transaction log and publishing them as events instead of adding another synchronous dependency.
A small interactive view of the same portfolio, shaped like an API response. A playful nod to the interfaces I build every day.
I'm interested in senior individual-contributor roles focused on backend engineering, distributed systems and AI-enabled products.