Shantanu Jumde

Built a next-generation AI-powered healthcare platform focused on reducing clinical workload through voice-driven automation.

• Designed and built a voice-based agentic workflow allowing doctors to operate the app using natural speech.Read Case Study
• Reduced 2–3 hours of manual discharge summary creation to ~10 minutes per patient.
• Built complete workflow end-to-end: frontend triggers, model orchestration, eval loops, step execution.
• Created highly modularised architecture, enabling deployment to Google Cloud, AWS, and On-prem environments.

  Created highly modularised architecture, so we were able to deploy it to GCP, AWS, On-prem, etc. to meet diverse client infrastructure needs.

  Created highly modularised architecture, so we were able to deploy it to GCP, AWS, On-prem, etc. to meet diverse client infrastructure needs.

• Tech stack: Next.js, TypeScript, Node.js, tRPC, Whisper, GPT-4, Langfuse, pgvector, Tailwind.

Led development across core consumer booking flows and internal sales automation systems.

• Impact: Delivered Reviews and MyBookings apps with complex pricing calculations, leading to a 30–50% increase in conversion across various geographies due to improved UX.
• VoiceIQ: Automated manual call verification, saving 6–7 hours/week per salesperson (1 full day/week), leading to increased productivity.

  Initially, the sales team had to manually verify each other's calls on 63 different parameters, spending 1 hour of their day on this task. This led to losing 6–7 hours/week per salesperson (1 day/week lost). We automated this end-to-end in just 4 weeks with AI-assisted delivery.

  Initially, the sales team had to manually verify each other's calls on 63 different parameters, spending 1 hour of their day on this task. This led to losing 6–7 hours/week per salesperson (1 day/week lost). We automated this end-to-end in just 4 weeks with AI-assisted delivery.

• Hackathon: Led a team to create a single-page checkout, reducing the process from 3 steps to 1, significantly improving user experience.

  Competed in a hackathon to improve checkout, creating a single-page checkout where 3 steps were reduced to 1, leading to a much better user experience.

  Competed in a hackathon to improve checkout, creating a single-page checkout where 3 steps were reduced to 1, leading to a much better user experience.

• Team Management: Led a team of 4 developers, collaborating closely with a Product Owner, Business Analyst, and Business Heads.
• Technical: Rebuilt slow Drupal-based booking experiences into modern, scalable React-Node.js (trpc/nestjs) micro-apps.
• VoiceIQ Pipeline: Designed complex AI pipeline using Gemini to transcribe, analyze, and score sales calls. Built on AWS Lambda + Step Functions with PII handling.

Modernized UI components and improved performance for booking websites.

• Promoted to Tech Lead for successfully handling a team of 4 and delivering complex applications (Checkout, Wishlist) as a junior engineer.
• Delivered Checkout and Wishlist apps with complex sharing features, reducing dependency on legacy Drupal software and manual sales processes.
• Optimization: Reduced payment and booking confirmation steps from 5–6 (Drupal) to 3 steps by implementing SQS requests to booking engines.

  Legacy Drupal systems took 5-6 steps to complete payment and confirm trips. We reduced this to 3 steps using SQS requests to booking engines (Softrip) and updating HubSpot to trigger email workflows.

  Legacy Drupal systems took 5-6 steps to complete payment and confirm trips. We reduced this to 3 steps using SQS requests to booking engines (Softrip) and updating HubSpot to trigger email workflows.

• Reporting System: Built a reporting system using SharePoint to provide actual numbers and accurate timestamps for debugging and analytics.

  Created a reporting system with SharePoint used by the sales team to get actual numbers, which later turned out to be very helpful for the business and development team for debugging numbers.

  Created a reporting system with SharePoint used by the sales team to get actual numbers, which later turned out to be very helpful for the business and development team for debugging numbers.

Contributed to a banking platform involving complex foreign exchange processing.

• Created Python-based CLI apps for calculations, consuming APIs, and updating data in the database.