AI & Backend Engineer · IIT Patna

Sujatha Karthikeyan

I build agentic AI systems — copilots, autonomous SDRs, and MCP integrations — and the backend infrastructure that runs them.

sujatha.ai · ask me anything
>
01 / About

I design and ship agentic AI systems end to end — from LangGraph pipelines and RAG to the FastAPI services and cloud infra behind them.

Currently an AI Engineer at Pulse, previously at IntoLeads. B.Tech in Information Technology from IIT Patna. I care about systems that are reliable, evaluable, and actually move a metric — not demos.

02 / Experience
Pulse
AI Engineer
Jun 2025 — Present
Analytical Copilot. Built an AI copilot for product teams with a custom eval system to benchmark response quality, plus compliant collateral generation under data-retention constraints.
MCP Server. Shipped Pulse's MCP server with a full OAuth connector flow, letting external AI clients interact with Pulse data programmatically.
Integrations. Extended Salesforce & HubSpot (parent-child account hierarchies, feedback sync), added Slack & Pendo as feedback sources, and built external APIs + a JPD downstream push to Jira Product Discovery.
Travel Automation POC. Led a TMC discovery session and built an agentic POC for end-to-end corporate travel — search, hold, book, human escalation, and cancellation.
IntoLeads
AI Engineer
Apr 2024 — Jun 2025
AI SDR. Owned core development of an AI Sales Development Representative on LangChain + LangGraph, cutting manual SDR workload by 40% through automated lead qualification, enrichment, and personalized outreach.
Agentic Workflows. Built human-in-the-loop pipelines for real-time lead verification, dynamic email generation, product-inquiry handling, and call scheduling.
Multi-Agent RAG. Built RAG workflows with LangChain, LangGraph, and pgvector to embed and retrieve sales docs for context-aware outreach, plus a LangGraph SQL Agent with dynamic few-shot injection from a vector store.
Backend & Infra. FastAPI backend serving AI workflows and a real-time analytics dashboard, deployed on AWS (S3, EC2, AppFlow).
03 / Selected Projects
NLP · EMBEDDINGS
Trending Keyword Dashboard

Text clustering with OpenAI embeddings + K-Means, surfaced through an interactive Streamlit dashboard.

Python · OpenAI · K-Means · Streamlit
DEEP LEARNING · LSTM
Air Pollution Forecasting

Bi-LSTM models with cyclical time encoding to forecast pollutant levels from meteorological, emission, and traffic data.

Python · Keras · TensorFlow · LSTM
ML · REGRESSION
Ski-Jump Scour Estimation

Ten M5P-based regression algorithms to estimate scour downstream of a ski-jump bucket.

Machine Learning · Regression
04 / Stack
LANGUAGES
JavaScriptPythonSQLBash
AI / FRAMEWORKS
LangChainLangGraphMCPFastAPINodeJSExpressJSMongoose
DATA / TOOLS
PostgreSQLMongoDBRedisDockerGitLangSmith
PLATFORMS
AWSGCPLinux
CERTIFICATIONS
Microsoft Certified: Azure AI FundamentalsApr 2025
LangGraph: Develop LLM-Powered AI Agents · UdemyNov 2024
PG Program in Securities Markets (Investment Advisory) · NISM2023–24