vinamr grover

Build end to end healthcare AI automations and workflows for Storyline's voice assistant, Cleo. Run production level workflow evaluations for MSK care, scoring against safety guardrails defined by US based medical professionals. Battle-test Cleo across multiple LLM providers (Claude / OpenAI / Gemini) on safety, helpfulness, tone, and clinical scope, all under HIPAA constraints.

Built custom event driven automations for the training and sales teams on top of the GHL CRM. Automated agent onboarding and rep workflows to keep high volume processes stable and resilient.

Wrote a service to monitor MongoDB health metrics across primary, secondary, and arbiter nodes. Handled core SRE/DevOps work: deploying CI/CD pipelines and provisioning infrastructure as code.

Orchestrated a distributed crawler on a fleet of t3.large spot instances to crawl millions of URLs for metadata extraction; output was zstd compressed parquet stored in S3. Pivoted the crawler from Python to Go; concurrency brought execution time down from hours to minutes. Chose ECR Serverless over Lambda for long running tasks; deployed containerized Lambdas as separate proxies to handle rate limiting. Pulled and classified data from 60k+ RSS feeds using LLMs; set up EventBridge schedules for interval and event driven pipelines. Built internal database auth and a pipeline for incremental backups and snapshots.

Provisioned and deployed full cloud infrastructure on Azure: databases, data warehouse, VMs, and web apps (internal + external facing). Published VM offers on the Azure marketplace used by clients globally. Built a Discord-based alerting and monitoring system for observability; deployed and served LLMs for internal tasks.

Being a Freelancer on Upwork, I used to extract data from multiple websites using Python Libraries such as requests, beautifulsoup and selenium. One prominent task was to scrape all the available short films on IMDB and leverage the Google Cloud's Youtube Data API to check their availability on Youtube.