The company behind SIMBA 2.0, the #1 TTS model in the world.

• Engineered a GCP-based migration pipeline to export and archive 1.54M+ Intercom chat support conversations and more than 1 TB of attachments to Google Cloud Storage, covering six years of support history across 304 weekly windows.

• Built a fault-tolerant, idempotent export workflow with streaming uploads, rate limiting, retries, and per-window reconciliation, achieving 304/304 verified windows with zero failed or truncated conversations.

• Migrated 1.05M+ attachments with 99.64% successfully archived, processing ~20K conversations/hour over 80+ hours of runtime, while avoiding buffering and eliminating redundant downloads through resumable exports.

Worked as a Software Engineer across selected engineering engagements, defense projects with the Indian Army, and open-source projects.

𝗦𝗲𝗹𝗲𝗰𝘁𝗲𝗱 𝗘𝗻𝗴𝗮𝗴𝗲𝗺𝗲𝗻𝘁 — 𝗭𝗲𝗻𝗮𝗰𝘁 𝗔𝗜

• Built backend services for mobile application asset management, implementing S3-based binary uploads and AWS DynamoDB metadata persistence, along with organization-level asset retrieval and reuse for automated testing workflows.

• Designed and implemented a scalable, decoupled Report Generation Service using a Fan-Out architecture: fetched test results from DynamoDB, formatted QA reports, and published events via SNS to SQS queues for async processing.

• Developed an independent Notification Service for asynchronous report delivery: AWS Lambda workers consumed SQS messages to send notifications via AWS SES (email) and Slack API, ensuring high availability and failure isolation.

• Implemented an LLM-agnostic agent layer supporting seamless switching between OpenAI, Groq, and Gemini through environment-based configuration, enabling zero-downtime provider changes and uninterrupted operations.

𝗗𝗲𝗳𝗲𝗻𝘀𝗲 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗖𝗼𝗻𝘁𝗿𝗮𝗰𝘁 — 𝗜𝗻𝗱𝗶𝗮𝗻 𝗔𝗿𝗺𝘆

• Engineered scalable backend services for a situational-awareness platform supporting defence operations, using FastAPI and PostgreSQL to ingest, process, and serve operational data to monitoring dashboards and command interfaces.

• Designed and implemented RESTful APIs with FastAPI and PostgreSQL for low-latency data access, using query optimization and strategic indexing alongside Celery and Redis for asynchronous processing of long-running tasks.

𝗜𝗻𝗱𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 & 𝗢𝗽𝗲𝗻 𝗦𝗼𝘂𝗿𝗰𝗲

• Contributed to open-source projects including Cortex, Prophetverse, and Open edX, implementing runtime feature discovery, Beta likelihood support for probabilistic forecasting, and technical-debt cleanup.

• Implemented end-to-end healthcare automation workflows, processing scanned documents and faxes with Tennr’s RaeLM to extract structured data and populate client EHR systems, improving accuracy and reducing manual effort.

• Engineered workflow optimizations that reduced approval and processing delays by 20% by improving triage rules, error-handling branches, and follow-up logic for missing or incomplete documents, accelerating reimbursement cycles.

• Designed workflow logic for document classification, data extraction, patient qualification, and insurance verification, ensuring accurate routing and reliable population of client EHR and billing systems.

• Monitored live automation pipelines across multiple healthcare clients and resolved workflow-level issues, configuration bugs, and edge-case failures, maintaining 99.5% uptime for all assigned accounts.

• Collaborated with operations and engineering teams to validate outputs, refine extraction rules, and improve system reliability, ensuring consistent performance for clinics, DMEs, and specialty practices.

• Delivered client-specific implementations for Neb Medicals, Aveanna, Gammie, Lymphacare, and other healthcare providers, tailoring automation logic to their documentation formats, reimbursement workflows, and operational needs.