rohney.io · Senior TPM at AWS

ROHNEY
PUSULURI

I build systems that outlast me, programs that run without me, and frameworks that every team after me just calls "the standard."

Technical Program Manager Software Engineer Forward Deployment AWS · Route 53 FedRAMP · IL4 · IL5 Distributed Systems Yoga Instructor
View LinkedIn See the work ↓
$4.4B
NZ Region
Sole TPM
60×
Perf. Gain
Consensus Algo
95%
Per-Region
Effort Reduction
3YR
Zero Breaches
FedRAMP
73%
MTTR
Reduction
8+
Global Regions
on My Framework
scroll to explore
Who I Am

Wired differently.
Useful everywhere.

I grew up with a bioscience teacher for a mom who made me deeply curious about how systems fail — and how people heal them. I wanted to be a doctor. Then I realized I could do the same thing for software infrastructure at AWS scale. I joined AWS as a software engineer and worked my way into owning programs that touch millions of endpoints, classified government cloud, and one very large $4.4B country-scale region launch.

I am not a TPM who manages Jira tickets. I write production code, design consensus algorithms, and build the tooling that makes everyone around me faster. I pick up what is about to fall before it hits the floor.

"I built RedSkull because I couldn't bear to see another new hire go through what I went through. I built Friday Recap because verbal standups are unreliable. I built the Tampermonkey tracker because I was on call and felt the pain myself. These weren't assigned. They were necessary."
01

Force Multiplier

I don't just do my job. I redesign the conditions so others do theirs better. Every framework I build gets inherited by the next team.

02

Calm Under Chaos

Yoga instructor. NCC-trained. Incident commander on 112-minute production outages affecting 6.85M endpoints. These things are related.

03

Teacher at Heart

I taught math and physics before tech. That instinct to transfer knowledge carefully shows up in onboarding programs, cross-team syncs, and DEI talks for AWS communities.

04

Structural Problem Solver

I look for root causes, not symptoms. The capacity audit that unblocked the NZ region build happened because I asked one question in a kickoff meeting that nobody else thought to ask.

Proof of Work

Not a resume.
A track record.

🌏
TPM · Geographic Expansion

AWS New Zealand Region Launch

Sole TPM on the $4.4B region build. No playbook. No prior solo-TPM precedent at this scale. I mapped 500+ host classes, audited every dependency within 48 hours, standardized 250+ CI/CD pipelines, and automated all 100 service components in Python. Shipped September 2025, zero SLA breaches.

95% per-region engineering effort reduction. 8+ global regions now built on this framework.
⚡
SDE · Distributed Systems

Consensus Algorithm: 112-Min Outage Resolved

A single Global Aggregator published bad data. Blast radius: 100%. 6.85M endpoints. No consensus mechanism existed. I evaluated Paxos, Raft, and majority voting, then built a 2-of-3 quorum design with graceful degradation. Delivered through 5-phase validation in 4 weeks. In production.

60× performance improvement (2.7B → 49M ops/cycle). 9× data consistency gain.
🔐
TPM · Security Program

FedRAMP: 3 Years, Zero Breaches

Led 10+ security campaigns across commercial, FedRAMP US Gov (IL4/IL5), and air-gapped classified environments. 1,000+ regulated endpoints. Built the execution framework from scratch: risk-based prioritization, escalation protocols, corrective action tracking, and SQL dashboards for monthly VP reviews.

71% audit prep reduction. 60% less vulnerability exposure. 83% fewer escalations.
📊
SDE · Observability

Health Analytics Platform

Built the entire stack from scratch: Java, Elasticsearch, Kibana, SQL. 10M+ daily metrics. 1M+ endpoints. I added whale customer monitoring with a custom Python scraper that fired alerts before enterprise issues became visible to customers. The team went from reactive firefighting to proactive prevention.

73% MTTR reduction. 40% of incidents caught before reaching customers.
🛡️
SDE · Security Engineering

Cryptographic Authentication System

Millions of health check packets were completely unauthenticated. A compromised monitoring node could inject arbitrary data with zero detection. I built a versioned cryptographic signing-key system in Java with region-isolated keys, integrated into the 6-thread UDP pipeline. Zero downtime across 38+ regions.

38+ global regions secured. Seamless key rotation, no coordinated downtime required.
🔧
FDE · Tooling & Automation

Tampermonkey Deployment Tracker

During live Sev2 incidents, finding which of 300+ pipelines caused a regression meant navigating multiple systems manually. I wrote a browser script that scraped 700+ deployment pages and injected the latest state directly into incident tickets. Nobody asked me to. I was on call, felt the pain, and built the fix.

40% reduction in Sev2 triage time. Zero infrastructure cost. One week to build.
Career Arc

How I got here.
The non-linear path.

2017 — 2018
Math & Physics Teacher
India

Before tech, I taught. This wasn't a detour. Every cross-team knowledge transfer, every AWS onboarding program I built, every VP briefing I simplified — this background is in all of it. Teaching forces you to find the explanation that actually lands, not the one that sounds impressive.

Knowledge Transfer Communication Curriculum Design
2018 — 2019
Full Stack Engineer + Scrum Master
Accenture · Hyderabad

Joined a Java high-availability trading platform. Sprint completion was erratic. I took on the Scrum Master role alongside engineering, introduced explicit dependency mapping at planning, and built a blocker escalation protocol. Sprint completion improved 39% across 3 releases. The instinct to identify a gap, own it without being asked, and deliver a measurable fix — that pattern has repeated itself ever since.

Java Full Stack Agile/Scrum Process Design
2019 — 2021
M.S. Engineering Science (Data Science)
State University of New York at Buffalo

Built predictive models on clinical health data, medical image classification with CNNs for cancer detection, and transaction fraud detection at scale. The medicine-adjacent work was deliberate. I still want to understand how systems fail and how to detect failure before it becomes catastrophic — whether that system is a human body or a distributed fleet of 6.85M health check endpoints.

Python Machine Learning CNNs · Keras SQL · R Clinical Data
2021 — Present
SDE → Senior Technical Program Manager
Amazon Web Services · Route 53 Health Checks · Seattle

Started writing production code. Built the consensus algorithm, the health analytics platform, the cryptographic authentication system. Then grew into owning programs at a scale where writing code became one tool among many. Sole TPM on the $4.4B New Zealand region build. FedRAMP program for 3 years, zero breaches. Founded RedSkull, built Friday Recap, wrote the Tampermonkey tracker. Programs I built now run regions I will never manage. That's the goal.

Java · Python · SQL Distributed Systems FedRAMP IL4/IL5 Program Management Incident Command Executive Comms
Engineering Projects

Code I shipped.
Problems I chose.

SDE · Distributed Systems

Quorum Consensus Algorithm

After a 112-minute production outage, I evaluated Paxos, Raft, and majority voting before choosing a 2-of-3 quorum design with graceful geographic degradation. Implemented in Java, integrated into the existing packet processing pipeline with stripe-level fault containment. Five-phase validation before production.

60×Faster
9×Consistency
4wkTo Prod
0Outages Since
JavaDistributed SystemsQuorum ConsensusPaxos/Raft AnalysisUDP Packet Processing
Context

Route 53 Health Checks monitors 6.85M endpoints across 64 global edge locations. One misconfigured Global Aggregator could propagate incorrect health data to the entire fleet. Blast radius was 100% before this fix. It is now single-source scoped.

2.7B ops/cycle → 49M ops/cycle
45% inconsistencies → under 5%
100% blast radius → single-source scope

SDE · Observability

Health Analytics Platform

Full stack: Java, Elasticsearch, Kibana, SQL. Custom Python scraper for whale customer monitoring. 5-min collection cycles, bulk processing 1,000 configs per batch.

10M+Daily Metrics
1M+Endpoints
JavaElasticsearchKibanaSQLPython
SDE · Security Engineering

Cryptographic Auth System

Versioned signing-key system in Java. Region-isolated keys. Integrated into 6-thread UDP pipeline. Fail-safe design — key retrieval failure never causes a service outage.

38+Regions
0Downtime
JavaCryptographyKey ManagementUDP
FDE · Tooling

Tampermonkey Deployment Tracker

Browser script. Scrapes 300+ pipelines, 700+ deployment pages, injects live state directly into Sev2 incident tickets. Zero infrastructure. Zero approval process. One week.

40%Triage Reduction
JavaScriptTampermonkeyWeb ScrapingIncident Management
FDE · Automation

Region Build Automation

Python scripts for dependency graph automation. Compressed 20-day manual region build process to 1 day across 100 service components and 3 partitions.

20×Faster
PythonCI/CDDependency GraphsAWS
Data Science

Medical Image Classification

CNN-based cancer detection on WDBC database. Classified FNA cells as Benign/Malignant with 97% false label reduction. Multi-layer neural network with Keras on Fashion-MNIST.

97%False Label Reduction
PythonCNNKerasTensorFlowClinical Data
Capability Matrix

One person.
Three fluencies.

SDE
Software Development Engineer
Java production code: consensus algorithm, cryptographic auth, health analytics platform
Python automation: region builds, dependency graphs, customer monitoring scripts
Distributed systems: quorum consensus, fault containment, UDP packet pipelines
Observability stack: Elasticsearch, Kibana, SQL dashboards, real-time telemetry
Security engineering: IMDSv2 migration, cryptographic signing, FedRAMP remediation
Full SDLC: CI/CD pipelines, Jenkins, Docker, GitHub, 250+ pipeline templates standardized
Machine learning: CNNs, logistic regression, clinical data, fraud detection
TPM
Technical Program Manager
$4.4B NZ region launch as sole TPM — zero SLA breaches, zero escalations
$10M+ cross-org reliability program: 50+ engineers, 3 organizations, 3+ years
FedRAMP program ownership: commercial, IL4/IL5, air-gapped classified — 3 years, zero breaches
VP and Director stakeholder management: SQL-driven KPI dashboards, monthly business reviews
Asana + Slack automation: ticket routing, SLA tracking, escalation triggers at scale
Risk-based prioritization, dependency graph analysis, OKR and milestone frameworks
Incident command: on-call lead for Sev1/Sev2 events on 6.85M-endpoint system
FDE
Forward Deployment Engineer
Customer obsession: built proactive whale-customer monitoring before anyone asked
On-site problem solving: identified and fixed Sev2 triage pain during live incidents
Cross-team leadership: trusted advisor to US Gov, IL4/IL5, air-gapped classified teams
Knowledge transfer: DEI talks, Women in STEM, Neurodiversity communities at AWS
RedSkull: built knowledge infrastructure for 30+ engineers across 4 teams
Friday Recap: redesigned team operating mechanism, now serves performance reviews
Capacity planning leadership: first cross-partition infrastructure inventory in org history
Outside the Terminal

The rest
of the stack.

I am certified yoga instructor who completed training in 2024 after nearly two decades of practice. The same intentionality that makes a yoga practice work — breath before reaction, stillness under pressure — is what I bring to incident command at 2am.

I paint. I skate. I grow plants. I trained myself to be ambidextrous because I was curious about how the brain adapts. I read Kahneman and Cialdini for fun — cognitive biases, cultural impact, influence at scale — and I use all of it in how I communicate tradeoffs to leadership.

My mother was a bioscience teacher who raised me to be curious, thoughtful, and inclusive. That shaped everything. I speak at AWS communities for Women in STEM, Neurodiversity, and persons with disabilities. I built RedSkull because I couldn't watch new hires struggle through the same painful onboarding I went through.

I care about sustainability, practice it actively, and believe that building systems that run without you is the most sustainable thing an engineer can do.

🧘Yoga Instructor
🎨Painting
🌱Plants
📚Kahneman · Cialdini · Goleman
🛹Skating
🔬Bioscience Roots
✍️Ambidextrous
🌍Sustainability
🎤DEI Speaker
What's next

Ready to build
something that lasts.

Looking for TPM II/III and Forward Deployment Engineer roles where I can get close to customers, build fast, and leave the team better than I found it.

Connect on LinkedIn