Making Ray Data best data processing engine for AI & Multi-modal data!

3P Fulfillment Integration

Led x-functional team spanning across a dozen orgs w/in Uber to build a new Personal Mobility product, integrating 3P Fulfillment Service Providers to be able to fulfill Riders transportation requests, while providing for Uber native experience.

First such integration was w/ Autocab (acq’d by Uber in 2020), which provides SaaS solution to taxi/PHV/limousine companies, providing them technological foundation to accept bookings, dispatch drivers, track availability, etc.

This new integration enabled Uber’s customers to request rides in the Uber app, via 3P Partners, in the areas where Uber’s primary Rideshare portfolio was not available.

My role was to TL the project, and as such

# Guided technical roadmap and overall execution of the project, to assure

convergence across all domains involved

## Setup venues and channels for effective collaboration environment

## Setup execution framework w/ clear x-team engagement model and clear OKRs

## Worked closely w/ Tech/Product/Legal/Ops orgs to clarify, refine and iterate on

product/legal/compliance requirements

## Engaged w/ Leadership stakeholders to communicate progress, manage

expectations and coordinate overall direction

# Designed E2E architecture, engaging with partner teams/orgs to collaborate on the integration points w/in their respective domains

# Coded major infrastructural components enabling 3P to integrate seamlessly

w/ Uber’s Fulfillment Platform

# Setup testing infrastructure/processes to enable teams to execute independently

avoiding stalls due to being blocked by another team’s progress

Some press on this:

Rebuilding Uber’s Fulfillment Platform (Project Horizon)

Next-gen of Fulfillment Platform for Uber’s scale built on top of the horizontally scalable storage solution (Cloud Spanner) supporting distributed transactions.

# Led various charters w/in the task force group of Engineers designing, experimenting,

prototyping and eventually onboarding onto the new Framework underlying

the new Platform

## Designed and implemented new asynchronous Application Framework leveraging

JVM, Netty, gRPC, Spring, etc. providing set of vital capabilities

out of the box (logging, metrics, tracing, HTTP/Thrift/gRPC clients, application

testing framework, etc), which

allowed to

### Save 1000s of engineer x hours, with widely reusable components and

repeatable flows

### 100s of engineers to successfully contribute to migration of the existing

product flows from previous Platform onto the new one

### Build a dozen of robust & reliable applications on top of it

### Influenced company-wide JVM-based framework

# Re-designed and implemented Asynchronous Spanner Client (not provided by

Google at the time), supporting

## Asynchronous operations execution (not supported by native client)

Intelligent lazy execution of operations analyzing data dependencies

between them, and allowing to

### Batch operations on the client level (multiple writes w/in the same

txn), and therefore

### Save on Uber <> GCP Network latency overhead (reducing # of

roundtrips), since all requests w/in Spanner txn have to be executed in

serial

# Led & executed multiple optimization projects targeting shared, low-level

infrastructure components widely used across the company

## Diagnosed, triaged, scoped and executed a series of substantial refactorings

of metrics reporting library (tally-java) to reduce excessive CPU & Memory

(GC) churn. As a result reduced library footprint w/in the application from 25%

of CPU utilization to ~5% (we emit a lot of metrics!).

# One of the top code contributors