Agentic AI Engineer · Bloomington, Indiana

I build AI agents that act.

I design multi-agent systems that move beyond chat — reasoning, using tools, and taking verified action in production environments.

Explore the proof
live_agent_trace.log
// Autonomous defense loop
objective: secure_without_breaking_users
01
Observeinspect live attack telemetry
RUNNING
02
Reasonpropose WAF rule changes
QUEUED
03
Actstage in COUNT mode
QUEUED
04
Verifymeasure + enforce safely
QUEUED
› Guardrails loaded in code
› Attack success: 67%
› False positives: 0.0%
Available
for the next
mission ↗
67→0%

Attack success eliminated across five autonomous defense rounds.

60%

Less manual processing through AI agents deployed across 7 IU campuses.

99.9%

Production uptime on an AWS-hosted platform built and shipped solo.

Selected systems / 2025–26

Proof, not
promises.

Every project closes the loop between intelligence and execution — with the metrics, guardrails, and production discipline to make agentic AI useful.

01 / FLAGSHIP

WAF Red-Team / Blue-Team Sandbox

Attacker and defender LLM agents contend over live AWS WAFv2 logic. The defender writes, stages, and verifies rules autonomously, while code-level safety gates protect real users.

Multi-agentAWS WAFv2Boto3Claude + GPTCloudWatch
↗ Launch live sandbox
Adversarial simulation verified outcome
67%R1
48%R2
31%R3
14%R4
0%R5
ROUND 01Initial exposure mapped — 67% attack success
02 / OPEN SOURCEAPR 2026

StandupBot MCP

An MIT-licensed MCP server that connects Claude directly to GitHub and turns 24 hours of commits, PRs, and reviews into a structured standup.

Node.jsMCPGitHub API
↗ View source
03 / MULTI-AGENTFEB 2026

Autonomous Route Optimizer

Gemini-powered agents coordinate 10+ stops in real time with perfect pickup-before-dropoff constraint adherence and voice-first interaction.

GeminiMaps APIWeb Speech
↗ Try it live
04 / CLOUD PLATFORMMAR 2025

CuraSure

A containerized healthcare platform with 30+ secure REST APIs, role-based journeys, 2FA, encryption at rest, and audit logging on AWS.

ReactGoAWSDocker
↗ View platform

Production track record

Built in the
real world.

From Mercedes-Benz scale to university-wide AI automation, I pair rapid experimentation with the reliability expected of production software.

May 2026 — Present
Bloomington, IN

Agentic AI Engineer

Indiana University

Deploying Copilot Studio agents for capital reporting across 7 campuses and piloting a 3-stage Claude code-interpreter pipeline that turns multi-day reconciliation into same-day analysis.

Nov 2024 — May 2026
Bloomington, IN

Project Intern

Indiana University

Solo-architected a production React, Node.js, Go, and MongoDB platform with Auth0 SSO on AWS, achieving 80% test coverage, zero post-release auth defects, and 99.9% uptime.

Aug 2022 — Jul 2024
Bengaluru, India

Software Engineer

Mercedes-Benz R&D India

Owned full-stack delivery for a production website, halving load time from 4s to 2s and sustaining 100% uptime. Built cross-platform products and lifted deployment success to 90%.

How I engineer

Agency with
accountability.

Guardrails belong in code.

Models can recommend; deterministic systems enforce. I design bounded tools, staged rollouts, permission gates, and explicit checks around every consequential action.

Evaluation is a product feature.

An agent is only useful when its outcome is measurable. I instrument task success, false positives, latency, and business impact from the beginning.

Prototype fast. Ship responsibly.

My foundation in full-stack and cloud engineering means experiments do not stop at the notebook. I can own the path from orchestration graph to dependable user experience.

Make complexity legible.

I translate agent traces and pipeline outputs into clear business conclusions — for the engineers operating the system and the leaders investing in it.

Let’s build what comes after chat

Need an AI engineer who can make agents useful?

Start a conversation ↗ View résumé