TeleCRM (recorded call) - MoM by email to customer

We are looking for an experienced Backend Developer to build a custom automation pipeline for our manufacturing and business operations. The goal is to automatically capture call recordings from our CRM (TeleCRM), transcribe them (handling mixed English and regional Indian languages), generate structured Minutes of Meetings (MoM) using an LLM, and auto-email the summary to the customer. Email service - Turbify, formerly yahoo small business The Workflow You Will Build Webhook Receiver: Create a robust API endpoint to handle incoming webhooks from TeleCRM when a call ends or a recording syncs. Capture the audio file URL, customer metadata, and email address. Multilingual ASR Pipeline: Pass the audio payload to a high-accuracy Speech-to-Text engine (preferably OpenAI Whisper Large-v3 or Google Cloud STT). The pipeline must robustly handle code-switching and mixed languages (English, Hindi, Marathi, Kannada, and Tamil) and output a unified English transcript or translation. LLM Summarization Engine: Pipe the transcript into an LLM (OpenAI API or Anthropic API) with engineered prompts to extract structured MoM (Discussion points, Decisions, and Action Items). Email Delivery System: Use Turbify, formerly yahoo small business to cleanly format and auto-fire the MoM to the client. Logging & Error Handling: Implement strict error handling (e.g., handling failed audio downloads, retries for API rate limits, logging webhook payloads). Technical Requirements Languages: Strong proficiency in Node.js (TypeScript) or Python. APIs & Integration: Extensive experience building webhook receivers, handling multipart form data/file downloads, and integrating third-party REST APIs. AI/ML Services: Direct experience working with OpenAI API (Whisper & GPT-4o) or Anthropic API (Claude). Proven ability to handle prompt engineering for structured JSON or Markdown text generation. Cloud & Deployment: Experience deploying production-ready APIs to cloud infrastructure (AWS Lambda/EC2, GCP, Docker, etc.). Localization Advantage: Familiarity with the nuances of Indian regional languages/accents and how ASR engines interpret them is a massive plus.