Senior Engineering Leader · Seattle

Sonu Gill

I lead engineering for consumer products — teams that move the business numbers, and now ship faster with AI.

Senior Director of Engineering, Member Success at Oportun. 17 years at Amazon before that. Based in Seattle.

Portrait of Sonu Gill
By the numbers

The proof, scannable in ten seconds.

Selected outcomes from recent years. Every metric below is verified.

+38%
App monthly active users, year over year in 2025
Oportun · Online Servicing
61% → 72.5%
Digital payment conversion in the app; 68% to 88% on the web
Oportun
$2M+
Online Servicing impact in 2025
Oportun
$350K+
Infrastructure savings in 2025
Oportun
12–16
Extra SDE-weeks per team, per quarter, from AI-assisted development
Oportun
$12M+ → $1M
Transcoding cost, brought down
Amazon · Photos
AI adoption at scale

AI that shows up in the roadmap, not just the demo.

In 2026 I moved every one of my teams at Oportun onto AI-assisted development, Codex included. That meant hands-on onboarding, biweekly knowledge shares, engineers demoing what they built, and a champion inside each team to keep it going.

Then we measured it. ROI reporting sits in our roadmap reviews next to everything else. The result: roughly 12–16 extra SDE-weeks per team every quarter, and about a 50% lift in velocity and code output.

AI is on the roadmap as a first-class line item — onboarding, knowledge transfer, champions, and ROI — not a one-off pilot.

Career

The arc and the scale.

Two roles, one through-line: engineering organizations that ship product and move the business.

Oportun Senior Director of Engineering, Member Success 2024 – Present

I lead Member Success engineering at Oportun, a regulated consumer lender: the Communications Platform and Comms Engine, Member Services, Onboarding, Savings and Online Servicing. I run it through directors and senior engineering managers and own priorities, quality bars and release gates across the portfolio.

  • App monthly active users up 38% year over year in 2025 across my Online Servicing org.
  • Lifted digital payment conversion from 61% to 72.5% in the app and 68% to 88% on the web, while moving the front end from Angular to React.
  • Delivered $2M+ in Online Servicing impact and $350K+ in infrastructure savings in 2025.
  • Brought tighter compliance and reliability to regulated member communications and payments.
  • Hiring: rebuilt hiring for our India team from scratch — interviewer training, a structured interview and debrief process, first-round automated coding interviews through a third-party service, and a mass-hiring process. Bar Raiser-style hiring without the name. It helped us scale by about 100 hires, and I taught key leaders to run interviews and debriefs so they could make fast decisions.
Amazon 17 years · Consumer Product Engineering 2006 – 2024

I started as a test engineer and grew into leading large consumer product orgs across Alexa, Amazon Photos and mobile.

  • Alexa Audio: led a 50-person org, including managers of managers, building music experiences on Echo Show, Fire TV, Alexa Auto, Echo Buds and mobile. Scaled from 1 manager and 19 engineers to 6 teams and 46 engineers, and opened the Vancouver office — a brand-new app development office, started from zero.
  • Cut the time to ship new features and screens from six months to six weeks with reusable components.
  • Raised employee satisfaction from 42% to 98%.
  • Amazon Photos: grew the computer vision charter from 12 to 23+ services, cut operating expense about 40%, and brought transcoding cost from $12M+ down to $1M.
  • Amazon Bar Raiser — 405+ interviews.
I still enjoy building

Trippits

Trippits (trippits.com) is a group trip-planning service I build and run on my own. A group votes on dates, location, stay, activities and travel until the plan is settled. It has an API designed for AI agents, and I use it as a live test bed for agent integrations.

Built & run solo API for AI agents Live test bed for agent integrations

podcast-skill Claude Code plugin · on GitHub · MIT

Given a topic, it researches with parallel sub-agents, re-checks the load-bearing claims against primary sources — verified vs. reported, in a sources.md alongside the episode — writes a two-host podcast script, then renders the audio and verifies it against the script. The voice and the transcription both run locally, and nothing is sent anywhere.

Topic → verified audio episode Claim-checked against sources Local voice + local audio QA

Beyond these, also built and shipped: finance-lens.com and investvsbuy.com.

Contact

Always glad to talk with engineering and product leaders and founders.

One conversation, via LinkedIn. No forms, no email gateways.

Connect on LinkedIn linkedin.com/in/sonugill