Dhruv Verma

Dhruv Verma

MS - Software Engg. @SJSU

Open to security & backend roles

dhruv12304@gmail.com

About

I'm Dhruv. Outside the keyboard you'll usually find me at a chessboard, across a table-tennis table, or in a pool — and on most nights, somewhere in a COD: Mobile lobby. I like going deep on things: conspiracies, evolution, physics, emerging tech, and drawing connections between events that look unrelated until they aren't.

Professionally I've spent my career in finance and banking, and I'm now turning that focus toward the security side of the domain — phishing, malware, ransomware, DDoS, insider threats, and man-in-the-middle attacks, the recurring class of problems that never quite go away. I've travelled across most of India and picked up a few languages along the way, which pairs well with the thing I love most outside work: long drives. Cars, open roads, no fixed destination. And if you ever want to argue something through, I'll happily play devil's advocate just to see where the seams are.

Selected work

2025–2026

Arbiter

Watching two AI agents argue about a decision

Automated systems make good decisions and bad ones, and from the outside they look identical. A number comes out, and someone has to sign off. Arbiter makes the argument visible instead. A scoring model picks the best delivery route on cost and time; a language model reviews that pick, argues with it in real time, and can override it. The operator watches both, and every decision is saved with the reasoning that produced it, replayable later.

The reviewer catches things the scoring model can't see, like a driver's free-text incident report, a zone advisory, or the fact that optimizing purely for cost quietly deprioritizes the same underserved neighborhoods every time.

Demonstrated on last-mile logistics, but the review layer doesn't care about the domain: swap the scoring function and the same audit applies to loan triage or ticket escalation.

Started as ALRO, a three-person team project at SJSU. I've kept building on it since. The audit trail, the fairness constraint, and the free-text signals came after.

  • Python
  • FastAPI
  • React
  • Docker
  • SQLite
  • SSE
  • LangChain
  • PyTorch
  • Pinecone
  • LLM APIs
Source

Several recent projects sit under NDA and aren't listed here.

Experience

  • 2025 — 2027 (expected)M.S. Software EngineeringSan Jose State UniversityGraduate coursework in distributed systems, cloud architecture, and AI engineering — extending eight years of fintech backend work.
    • Focus areas: distributed systems, scalable cloud architecture, applied machine learning, and software engineering process.
    • Capstone and side projects exploring agentic LLM systems, retrieval pipelines, and event-driven backend design.
  • Sep 2021 — Jul 2025Senior Software EngineerBetter MortgageScaled Tinman, Better's core loan origination platform — high-throughput Java/Python microservices that cut API latency by 20%.
    • Designed and scaled Tinman, Better's core loan origination platform, by developing high-throughput microservices in Java and Python that reduced API latency by 20%.
    • Engineered RESTful APIs and workflow automation using Java Spring Boot and PostgreSQL for complex financial computations (DTI and APR), ensuring seamless integration across partner and compliance systems.
    • Built automated data validation pipelines in Python and SQL to detect underwriting anomalies, reducing manual compliance review time by 15% and ensuring high data integrity.
    • Collaborated with infrastructure and backend teams to optimize services on AWS, ensuring high availability, system reliability, and scalable architecture across the platform.
  • Aug 2019 — Sep 2021Software EngineerXL Dynamics India Private LimitedRefactored a legacy mortgage servicing system into modular Spring Boot, improving DB performance 18% and automating compliance workflows.
    • Refactored a legacy mortgage servicing system into a modular Java Spring Boot architecture, optimizing PostgreSQL database queries to improve performance by 18%.
    • Automated document validation and regulatory checks by writing backend automation scripts in Python, increasing compliance processing speed.
    • Streamlined the internal compliance portal using HTML and JavaScript, reducing loan processing turnaround time by 20% for operational teams.
  • Nov 2018 — Jun 2019Business AnalystKhurana & Khurana IP LawTechnical bridge between patent attorneys and startup clients, translating software architectures into actionable IP strategy.
    • Served as the primary technical bridge between patent attorneys and startup clients, translating complex software architectures into actionable IP strategy.
    • Conducted technical due diligence on software assets for early-stage technology companies, informing defensibility assessments and patent filing decisions.
  • Mar 2015 — Nov 2017AnalystXL Dynamics India Private LimitedBuilt compliance-embedded backend services and ran end-to-end UAT for production mortgage releases.
    • Built backend services that embedded compliance logic directly into the core mortgage platform, ensuring regulatory requirements were enforced at the system level.
    • Executed end-to-end system and UAT testing using JUnit and Postman, validating financial data accuracy and workflow correctness before each production release.
  • 2011 — 2015B.Tech, Computer Science & EngineeringBBDEC Lucknow · AKTUComputer science fundamentals — data structures, operating systems, databases, networks — and early Java/web backend projects.
    • Core coursework across data structures, algorithms, operating systems, DBMS, computer networks, and software engineering.
    • Capstone and internship projects in Java and web backends that laid the foundation for a career in fintech infrastructure.

Writing

Contact

Reach out for security and backend roles, collaborations, or a good argument.