Resume

• •Migrated uyir.ai from OTP-based sign-in to ZITADEL OIDC and synchronized existing user accounts.
• •Built therapist booking with slot validation, Razorpay payment capture, and LiveKit room creation.
• •Processed signed LiveKit webhooks to track session attendance and prevent duplicate event handling.
• •Built text- and voice-based AI journaling with Tiptap inline editing and encrypted storage; blocked journal visualization when crisis content was detected.
• •Used presigned AWS S3 uploads for therapist media and knowledge-base files, keeping uploads off the application server.
• •Added offline states and global error boundaries to the Next.js application to handle network and rendering failures.
• •Automated deployments to a DigitalOcean development environment using GitHub Actions, Docker, and Alembic migrations.

• •Built the backend for a call automation SaaS using Go and Python microservices.
• •Enforced tenant isolation with ZITADEL authorization and PostgreSQL row-level security.
• •Configured APISIX to validate JWTs, rate-limit public routes, strip untrusted headers, and keep internal endpoints private.
• •Built an asynchronous document-ingestion pipeline with GCS and JetStream, then indexed Gemini/Vertex AI embeddings in Qdrant for retrieval during calls.
• •Built a background provisioner that initializes new organizations across services and retries failed steps independently.
• •Built LiveKit agents for inbound and outbound calls using Gemini, ElevenLabs, Cartesia, and Sarvam; captured recordings, transcripts, extracted fields, and usage for each call.

Sol, Smart Home Ecosystem – Go, Python, PostgreSQL, Redis, MQTT, LiveKit, Azure OpenAI

• •Built the Go backend for managing homes, rooms, members, appliances, and device state with PostgreSQL (TimescaleDB) and Redis.
• •Built a Python voice assistant with LiveKit and Azure OpenAI for natural-language device control.
• •Secured the device platform with ZITADEL authentication, room-level access controls, and mTLS-authenticated MQTT connections.
• •Tracked device commands with MQTT acknowledgements and control logs so the backend could confirm their outcome.
• •Built OTA firmware updates with mTLS-protected downloads, safe retries, and device configuration preserved across updates.

Zylert, AI Notification Triage App – Kotlin, Android SDK, Node.js, PostgreSQL, JSON/Zod, REST APIs

• •Built an Android app and Node.js backend that uses AI to classify captured notifications as important, moderate, or unwanted.
• •Added AI-generated notification summaries and a chatbot for searching and asking questions about past alerts.
• •Validated API payloads with Zod and stored notification history in PostgreSQL.