https://linkedin.com/in/jatinsonavane | Seattle, WA
Full-stack engineering leader with deep experience building scalable, production-grade systems. Worked in regulated financial environments and prop-tech. Passionate about fast-paced product delivery via mission-driven teams in high autonomy environments.
Growth | Cross-Team Adoption | Distributed Systems | Java | Backend APIs | Team-Building
Present | Vue, Phoenix+Elixir, Postgres, S3
Designing and building a real-time group communication platform designed from the ground up to preserve one's attention.
Jan 2024 – Jun 2025 | Databricks
Led two Growth engineering teams running experiments on Confluence and Jira.
Ensured multiple experiment streams were on track for delivery, followed best practices and provided as-needed support to engineers and product managers.
Translated ambiguous ideas into concrete experiments via SQL analysis on data lakes.
Drove cross-team adoption of internal platform improvements; resolved roadmap dependencies across engineering, data science, and product.
Mentored engineers; participated in performance calibration for 100+ engineers semi-annually.
Feb 2022 – Jul 2023 | Vue, Node, Postgres, Kafka
As a member of the leadership team, I shaped the teams that built Clarity, a cloud-based portfolio review platform that provides Capital Group Mutual Fund Managers with insight into their holdings, investment performance, and their equity trades. Clarity has an ETL to gather data nightly, uses Postgres RDS, a Redis cache, a Node back-end and a Vue front-end. It also consumes streaming equity trade data from Kafka via a C# service. All these components run on AWS.
Architected and built a near-real-time equity-trade view into Clarity (ingesting ~500K events daily).
Helped restructure a 24-engineer team into 3 focused teams; hired 8 engineers; stayed hands-on as manager of the streaming trade data team.
Jan 2014 – Feb 2022 | Java, React, SQL Server
2021: Senior Engineering Manager 2020: Principal Engineer 2017: Senior Engineer 2015: Engineering Manager 2014: Senior Engineer
ZO Auth Framework (internal platform): Designed and delivered a centralized OAuth2/OIDC auth platform adopted as the shared auth layer across over 10 engineering teams — designed for internal developer experience, with onboarding docs, migration guides, and a dedicated support group and channel. [Java, Redis, Terraform, Kong, OAuth2/OIDC]
ZO Workflow: Designed and built a Flowable-based event-driven workflow platform for Zillow Offers' iBuying operations. [Java, SQL Server]
Underwriting Platform: As a manager, I formed and led the Zillow Offers Underwriting Platform team in building web apps and services that determined the fees to charge home buyers. We worked on projects to reduce fee-generation time from 2 days (manual process involving a 30 MB Excel file) to 10 seconds (using ML models). [React, Java, S3, DynamoDB]
Seller Marketplace: Managed team working on various features for home sellers. Oversaw a monolith-to-microservices migration for seller services; grew team from 2 to 8 engineers. [Java, Python Flask, MongoDB, SQL Server]
Pricing KML Service: Built a shared home geo-location microservice in a weekend — replaced a 6-month roadmap estimate that had a dependency on another team.
Reduced company-wide CI pipeline from 3 hours to 8 minutes through a voluntary cross-team engineering initiative.
Jan 2006 – Jan 2014 | Slang, SecDB, Java, Sybase, DB2
2009: Vice President 2008: Senior Analyst/Developer 2006: Analyst/Developer
Supported the Fixed Income Structured Credit Derivatives trading operations business. Projects ranged from processing new financial instruments to reducing operational risk.
Credit Event Desktop: Six of us across New York, London and Bangalore wrote a distributed application to help trading and risk operations manage various bankruptcy workflows across domains like payment handling and confirmations. The UI was C#, the middle-tier was Java, and various API services were Slang+SecDB. [Java, C#, DB2, Slang, SecDB]
Credit Event Risk Report: This was a solo project where I refactored legacy scripts with poor readability that handled the impacts of potential credit events (bankruptcies) in companies we traded. The refactor caught and fixed uncaught bugs, and enabled scale by running on a machine pool. This was noteworthy as it improved Goldman's ability to manage its trade positions nimbly and accurately in the wake of the Global Financial Crisis of 2008-2009.
External Conversation Manager (ECM) Refactor: ECM was an existing Java workflow application (using Drools Flow) which helped our trading operations team coordinate between internal trade booking channels and various downstream teams. This application had become very difficult to enhance. Using TDD, I re-implemented it in 2 weeks, leading to tested and readable code. [Java, DB2, EMS]
Event Manager: We had a large number of systems supporting a variety of post-trade activities. Many of these raised alerts with either operations (via email) or tech support teams with the tools the support teams use. The Event Manager was designed as a Java middle tier service that accepts alerts from various systems, persists alert data, and routes alerts based on static rules to various different recipients. The development was done in a team of 3. [Java, DB2, SecDB]
Many projects to enable the downstream processing of new credit derivatives financial instruments. [Slang, SecDB]
MS Computer Science — Binghamton University, NY (GPA 4.0) Thesis: Generating Relevant Terms for Concepts in a Concept Hierarchy — word sense disambiguation at 90% precision over a 460K-concept hierarchy. Statistical ranking (ATF×ICF) combined with semantic enrichment via WordNet traversal.
Hobby projects feed curiosity. Here are some recent highlights:
Crickets (https://crickets.fly.dev): Chat with your friends without messages being stored in any data stores or logs. This is especially easy with Elixir's message-based inter-process-communication capability. I made this to learn Elixir, a functional language. [Elixir, Phoenix LiveView, no databases!]
Snake (https://snake.fly.dev): You'd write a game like Snake on the client side. But I wanted to see how fast Phoenix LiveView would be when implementing game logic on the server side. Not too bad! [Elixir, Phoenix LiveView]
Strawman (https://www.strawman.fyi): In 2023, I was worried that ChatGPT would make it so easy to get information that kids would no longer need to think critically to solve problems. This was a pretty tacky early attempt to use AI offline to generate logical fallacy examples that might make them easy to learn. Potentially, this could evolve into an AI-based interactive teaching course. [React, Node, Postgres]
OAuth2 Authorization library (https://github.com/spacecricket/authorize): A Java library that enables intuitive fine-grained claim-based authorization directly via Spring Boot controller method annotation. Its contribution is to keep authorization concerns as close to the endpoint as possible.
Table Tennis League Dashboard (https://www.thom.so/leagues): A vibe-coded dashboard for my son to see details about his performance in table tennis leagues. Contains play-by-play match recording as well as note-keeping. I built this to experience vibe-coding. [Next.js React, Postgres]
Languages: Java, JavaScript/TypeScript, Node, Python Platform & Ecosystem: Internal SDK design, experimentation platform patterns, cross-team adoption, shared API design, microservices Distributed Systems: Kafka, WebSockets, Redis pub/sub, event-driven architecture Data: Postgres, SQL Server, DynamoDB, MongoDB, Redis, S3 Auth & Identity: OAuth2, OIDC, token lifecycle management, claim-based authorization Frontend: Vue, React