Sarvottam Bhagat

Developed a Text-to-SQL app that allows users to interact with their database using natural language prompt. Automated testing with Playwright MCP Server, prompt-based script generation, cutting manual effort by 80%.

R&D Intern at Ellucian (2025-01 – 2025-04)

• Developed a Text-to-SQL app that allows users to interact with their database using natural language prompt.
• Automated testing with Playwright MCP Server, prompt-based script generation, cutting manual effort by 80%.

AI Engineer at TheAgentic (2025-07 – Present)

• Led AI equity tax platform (41K LOC) outperforming ChatGPT/Claude via hybrid architecture: GPT-4o intent mapping + 102 deterministic Python functions, achieving 100% accuracy on financial calculations.
• Architected 3-agent pipeline (Scenario Mapper → Argument Mapper → Executor) routing queries to 14 tax strategies across ISO/NSO/RSU/ESPP, covering 50-state multi-year scenarios.
• Built dual-mode system: scenario calculator + RAG chat (5-stage Qdrant) processing 150+ IRC documents with sub-3s response times and real-time citations.
• Deployed FastAPI microservices with 30 PostgreSQL migrations, 7 API integrations, async document processing (5-25 concurrent ops), Docker orchestration, IRC compliance, tax constants, and 35 test suites.
• Built AI sales assistant for founders automating email drafting, meeting scheduling, and CRM updates via LangGraph with 8 specialized agents (Gmail, Calendar, HubSpot, Slack, Sheets, Research) and PostgreSQL state persistence.
• Architected OAuth integration layer via Composio connecting 6 services (Gmail, Google Calendar, HubSpot, Slack, Google Sheets, Drive) with approval-flow pattern and user-scoped entity isolation.
• Implemented AI sales coach with MEDDIC deal qualification, DISC personality profiling, RAG document Q&A (pgvector), and adaptive onboarding for personalized coaching.

Bachelor of Science in Information Science and Technology – CMR Institute of Technology (2021 – 2025)