Pravin

AI Software Developer · Agentic Workflows

I build agentic AI systems that ship.

Ask the RAG Waiter anything about my work — it explains the hard stuff with restaurant-grade analogies.

What I do

The Technical Architect

Agentic AI Development

Multi-step agents that plan, call enterprise tools, and verify their own work — from prototype to production.

Scalable Cloud Infrastructure

AWS + Azure architectures designed for elasticity: containers, queues, and observability from day one.

Latency Obsession

Every millisecond audited — streaming responses, warm paths, and caching until the wait disappears.

Selected work

Projects

Enterprise Agentic ChatbotFlagship

→

A multi-service AI agent inside MS Teams: leaves, payslips, and ServiceNow tickets handled in plain language with full read/write/update capability.

Azure Bot ServiceMS TeamsNode.jsServiceNowHRMSAWS

OpenAI Realtime Voice Bot

→

Ultra-low latency AI voice assistant for a Hospital using OpenAI's Realtime API, Twilio, and Supabase. This system replaces traditional telephone operators with an intelligent AI that can understand, route calls, and provide information with ~500-800ms latency.

Node.jsExpressReact 18ViteTailwind CSSOpenAI gpt-4o-realtimeTwilio Voice APISupabase (PostgreSQL)WebSocketsWebRTCG.711 μ-law (8kHz) for phonePCM16 for web

MediScribe-AI | Ambient Clinical Intelligence Platform

→

Developed a full-stack healthcare application enabling clinicians to stream ambient audio during patient consultations for real-time medical documentation. The mobile app captures audio, streams to an API server, processes transcriptions via AI, and auto-fills EHR forms on a web dashboard. Implements real-time audio processing, AI-powered medical text extraction, and secure pairing mechanisms. Key Features: - Real-time audio transcription and medical documentation extraction - Live transcript streaming with silence filtering for optimal bandwidth usage - AI-powered EHR form auto-fill with manual edit tracking - Secure OTP-based device pairing - Responsive web dashboard and native mobile interface

React Native (Expo SDK 54)React + ViteNode.js + Socket.ioPython FastAPIOpenAI API (WhisperGPT-4)TypeScriptDocker

Everyday tools

Node.jsTypeScriptNext.jsReactPythonOpenAI APIAgentic WorkflowsAzure Bot ServiceAWSDockerSupabasePostgreSQLpgvectorTailwind CSSCI/CD

Flagship project

Enterprise Agentic Chatbot

A multi-service AI agent living inside MS Teams: employees apply for leaves, fetch payslips, and manage ServiceNow tickets in plain language — with full read, write, and update capability.

Channel

MS Teams

Employee-facing conversational surface

Gateway

Azure Bot Service

Secure gateway, identity & message routing

Bot Framework activity▾
Agent Core

Agentic Orchestrator — Node.js (AWS)

Intent planning, tool selection, and multi-step execution. An LLM decides which enterprise tool to call, validates the result, and composes the reply.

Tool calls (read / write / update)▾
Tool

ServiceNow

Ticketing — create, query, and update incidents from chat

Tool

HRMS

Leaves & payslips — apply, check balance, fetch documents

Traces, tokens & latency▾
Telemetry

Supabase Observability

Every turn logged with token count and latency — the same pattern powering this portfolio's admin dashboard

Where I've been

Experience

  1. July 2026 - Present

    AI Software Developer · Somvanshi Global Technologies Pvt. Ltd

    Architecting Solutions that Scale

  2. Dec 2025 - July 2026

    Jr AI Software Developer · Somvanshi Technologies Pvt. Ltd

    Developing AI Based Solutions in Healthcare, Finance and EdTech Sectors and Deploying End to End AI Devlopment Models.

    • —Cut median response latency with streaming + warm paths.

  3. June 2025 - Dec - 2025

    Technology Intern · Somvanshi Technologies Pvt. Ltd

    As a Tech Intern at Somvanshi Technologies, I was contributing to real-world projects in the domains of web development and artificial intelligence.

    • —Worked on AI-driven applications involving Computer Vision, LLMs, and Transformer models.

Publications

Research

PublishedIJFMR2026

MediScribe AI: Ambient Clinical Intelligence for Automated Electronic Healthcare Record Documentation

Mr. Pravin Dinkar Paithankar · Dr. Radha S. Shirbhate · Mr. Adarsh Dhanraj Wankhede · Ms. Shraddha Balasaheb Labade

Clinician burnout caused by extensive electronic health record (EHR) documentation is a significant issue, frequently diminishing the time allocated for patient engagement. To solve this problem, we developed MediScribe AI, an ambient clinical intelligence system that uses speech recognition, speaker diarization, and clinical entity extraction to automate medical documentation. The system uses OpenAI Whisper, MedCAT, ClinicalBERT, and GPT-4o to generate structured SOAP notes and enable EHR updates in real time. In this study, we examined 29 recent publications (2024–2025) and found that ambient AI scribes can substantially decrease documentation time and cognitive load. However, errors in speech recognition and named-entity recognition (NER) still require human review. MediScribe AI aims to achieve high accuracy, low latency, and strong usability, making it suitable for healthcare settings where resources are limited.

Read the paper ↗PDF ⤓

Open source

On GitHub

pravinpaithankar's GitHub contribution chart
Full profile on GitHub →

Notes & essays

Writing

All posts →

July 10, 2026

Solving the Latency vs Security tradeOff in realTime AI Applications

Discover how a 'Stream Bridge' architecture using Node.js, WebRTC, and WebSockets reduces AI voice assistant latency to under 1 second, enabling secure, interruptible, and human-like conversations.

Say hello

Whether it's an agentic system or a chess opening, I'm always up for a conversation.

Share a phone number or an email (at least one) so I can get back to you.

Elsewhere

GitHubgithub.com/pravinpaithankar→LinkedInwww.linkedin.com/in/pravinpaithankar→Instagramwww.instagram.com/pravin.ai→Google Developersme.developers.google.com/u/pravinpaithankar→