Engineering Leadership — Systems, Risk, Scale
Kaustubha Amaravadi
I build and turn around engineering teams — and the real-time systems they run — for problems where being wrong is expensive.
Portland, OR · Open to Engineering Manager / SDM roles — Portland or remote
Three teams, one operating principle: catch it before it costs anything. Founded two from scratch; inherited and rebuilt the third.
Systems built
Three teams, three production systems
Two started as a blank page — no team, no system, no roadmap. The third was an inherited team on a legacy system that needed both a rebuild and a reset. All three shipped into production at Amazon's scale.
Amazon
Concessions Risk Platform
Inherited team of 4 · grew to 9 · rebuilt the platform
Inherited a 4-engineer team running an aging, legacy concessions system — and a team that needed real leadership alignment as much as a technical rebuild. Replaced the old system with a new real-time, distributed risk and entitlement platform and grew the team to 9 (US and India). Delivered $100M+ in annual savings.
Amazon
Customer Contact Automation
8 SDEs · built from the ground up
Founded and scaled a new team to build AI-driven automation that handled abuse-flagged customer contacts directly — freeing up human agents to focus on the highest-risk cases instead of routine ones.
Amazon
Delivery Risk Scoring & Treatment
6 SDEs · built from the ground up
Real-time, checkout-time risk scoring that didn't just flag risk — it triggered risk-commensurate delivery treatments, like requiring a signature or OTP on delivery, gating high-risk deliveries at enterprise transaction volume.
Experience
Twelve years, three companies, one throughline
Distributed systems and risk-critical infrastructure, from IC to team-builder.
Software Development Manager
Amazon
Founded two engineering teams from zero (AI-driven automation for abuse-flagged customer contacts, freeing human agents for the highest-risk cases; delivery risk scoring & treatment) and inherited and rebuilt a third — growing an existing 4-engineer team to 9 (US and India) while replacing its legacy platform with a real-time risk and entitlement system — $100M+ in documented annual impact.
Software Developer
Amazon
Built core checkout services to proactively assess delivery failure risk, contributed to Amazon IoT device software, and designed a C-based service between OS and application layers.
Software Developer
GAP Inc.
Led implementation of a new Warehouse Management System (Manhattan WMOS), managing a cross-functional team of 4 engineers plus 5 contractors — owning hiring, budget and scope planning, and performance reviews for the contractor team.
Member Technical Staff
Oracle
Automated and optimized ETL testing using Oracle Data Integrator, contributing to Oracle BI's Student Information Analytics module.
Now
What I'm building outside the day job
Between roles, I keep shipping — smaller scale, same instinct for zero-to-one.
evchargesavings.com — EV Charging Cost Calculator & Ownership Hub
A cost calculator and ownership resource with zip-based geolocation, state electricity rates, and 30+ EV models — built solo from concept through deployment.
Agentic investing system
A Robinhood MCP integration with an autonomous portfolio-analysis operating prompt — daily analysis, human-gated trade execution.
Fractional engineering leadership
Advisory work in risk and fraud systems, and engineering org audits — drawing directly on the abuse-detection and trust-and-safety background above.
A point of view
AI is the calculator moment, not the disruption
Calculators didn't stop us from understanding math — they made it accessible. Suddenly everyone could do complex arithmetic in seconds, no theory required. Computers pushed that further: thousands of hours of work compressed into tens, using code. But the cost of entry stayed high. You still had to learn to program, understand systems, think like an engineer before you could build anything real.
AI is removing that gate. Anyone with a computer and an idea can now perform like an SDE, with almost no barrier to entry. A real MVP in hours instead of months. People who assumed they'd never be able to build a product now have the tools to just try.
For engineering managers, this changes what's worth hiring and leading for. When anyone can generate code, the scarce skill isn't typing it — it's specifying the problem precisely, judging whether the output is actually right, and knowing when to override the machine. That's the muscle I look for when I hire now, and the one I build deliberately in the teams I lead.
EVIDENCE, NOT OPINION
evchargesavings.com is the proof: an EV charging cost calculator and ownership hub with zip-based geolocation, live state electricity rates, and 30+ EV models — built solo, from domain and architecture through a full editorial design system, using AI-native tools the whole way. What would have taken a small team a quarter to scope and ship, I built and iterated alone.
Contact
Let's talk
Currently exploring Engineering Manager / SDM roles — Portland-area or remote. I build the systems that can't afford to be wrong — and the teams that run them. If that's what you need, let's talk.