Hi, I'm Sharath 👋

I build AI agents that ship to production.

A Full-Stack AI Engineer with 2.5+ years building production Agentic and Generative AI systems — multi-agent orchestration, RAG pipelines, and the full stack underneath them, in healthcare and enterprise settings.

AI Engineer2.5+ yrs full-stack
R1 RCMcurrent role
Fairfax, VAMS · GMU
LangChain + RAGagents are my love language
drag the notes around!
a bit about me

I make agents that earn their keep.

For 2.5+ years I've architected and deployed production-grade Agentic and Generative AI systems across healthcare and enterprise domains — the kind that process real transactions, not just demos.

I'm comfortable designing a multi-agent orchestration graph in the morning and tuning a RAG retrieval pipeline by afternoon. Right now I'm at R1 RCM, where my system autonomously processes 100K+ healthcare claims a day and cut decision cycles from 48 hours to 2.

My favourite part? The moment an agent that used to need a human for every edge case starts handling them on its own — safely, and with a full audit trail.

The quick facts

RoleFull-Stack AI Engineer @ R1 RCM
Based inFairfax, Virginia
StudiedMS Information Systems, GMU
Works inLangChain, LangGraph, FastAPI
Fluent inPython · TypeScript · SQL

Curious what this kind of role involves? Try my free learning module for AI Full-Stack and Forward Deployed Engineers.

what's in my toolkit

Everything I bring to the table.

↓ grouped like a real affinity map — drag the notes if you like

Languages & Frameworks

PythonTypeScriptFastAPIReact / Next.jsMicroservices

Agentic AI & LLM

LangChainLangGraphClaude APIRAGPrompt Engineering

Data & Persistence

PostgreSQLPineconeFAISSVector DBsRedis

MLOps & Cloud

DockerKubernetesAWSLangSmithTerraform
impact, in numbers

What the agents actually did.

100K+
healthcare claims processed autonomously, every day
65%
reduction in manual intervention on claims routing
62→89%
claims accuracy, after rebuilding the RAG pipeline
48h→2h
decision cycle time, with multi-agent orchestration
stuff I've built on the side

Personal projects, shipped for real.

↓ smaller in scope than the day job, built solo end to end

Split It

A Splitwise-style expense splitter — group balances, PDF/Excel exports, and dashboards that actually make sense.

PythonStreamlitPandas
MoodLens

A mood journal that listens — Whisper transcribes your voice, RoBERTa reads the emotion, and it tracks your streaks.

HuggingFaceWhisperStreamlit
LedgerLens

A finance dashboard that turns a boring ledger into interactive charts you can actually explore in real time.

PythonStreamlitPlotly
words, not just code

I explain AI in plain English.

AI in Plain English · MediumWhy AI Benchmark Scores Mean Almost NothingRead the article → AI in Plain English · MediumWhy Every Big Company Banned ChatGPT, Then Built Its OwnRead the article → AI in Plain English · MediumWhy AI Confidently Makes Things UpRead the article → AI in Plain English · MediumWhy Does a Chatbot Need a Nuclear Power Plant?Read the article →
More on Medium →
how I got here

The path so far.

Mar 2025 — now
Full-Stack AI Engineer
R1 RCM · Chicago, IL
LangChainLangGraphClaude APIRAGHIPAA
Feb 2022 — Jul 2023
Software Engineer
Highradius · Hyderabad, India
RPAOCRPythonAWSVue.js
2023 — 2025
MS, Information Systems
George Mason University · Fairfax, VA
Education
BTech, Computer Science & Engineering
JNTU · Hyderabad, India

What each role was really about

R1
100K+ claims/day, autonomously
Multi-agent orchestration · planner + executor
RAG
500K+ policies, hybrid search
Retrieval accuracy 78% → 94%
Ops
MLflow · LangSmith · Langfuse
Versioning, observability, A/B testing
HR
50+ Fortune 500 clients
RPA/OCR automation · 78% less manual entry
let's build something that ships

Got an agent that needs
to earn its keep?

Open to Full-Stack AI Engineer roles — agentic systems, RAG pipelines, and the production infrastructure underneath them. I'm quick to reply.