Resume · Madhur Mehta

Madhur Mehta

Vel Labs

Bengaluru, India

Forward Deployed Engineer

Feb 2026 – Present

Core developer of Sola, an AI companion app on a single Kotlin Multiplatform and Compose Multiplatform codebase shipping to Android and iOS (Decompose, Koin, Ktor).
Owned the real-time 3D Unity companion via Unity as a Library: built the KMP bridge module, a versioned message contract, thermal-aware frame rate and Sentry telemetry; removed a 3.5 s freeze and raised fps from 21 to 29 on low-end devices.
Set up performance engineering with Macrobenchmark, Baseline Profiles and Perfetto KPI budgets, cutting cold-start time by about 25% on average.
Closed 22 production crash, ANR and hang issues through Sentry root-cause analysis, moving SDK init and dependency injection off the main thread.
Built chat and WebRTC voice calls: paging, optimistic updates, swipe-to-reply, realtime reactions (Ably) and a shared-element photo viewer.
Shipped ringing reminder calls (Android Telecom), FCM push with deep links, RevenueCat and UPI payments, and funnel analytics.

Brick and Bolt

Bengaluru, India

Software Development Engineer - I (Android)

Nov 2024 – Jan 2026

Built a Server-Driven UI referral flow for the Customer and Professional Android apps, streamlining lead generation and conversion tracking.
Integrated the Google Play Integrity API to block unauthorized installs and rooted devices.
Built team management with role assignment in the contractor app, reducing manual coordination by about 25%.
Built Server-Driven UI components for the Customer app and a Flutter internal dashboard, enabling UI updates without app releases.

Pvot Phone & PvotLib | Kotlin, Jetpack Compose, Dagger/KSP

Moving an open-source AOSP dialer to Jetpack Compose: migrated 10 screens, in-call UI and 8 dialogs, removed 34k lines of dead code, and added caller screening and missed-call reminders.
Contributed about 35 components and a glass surface system (WCAG 4.5:1 contrast) to the PvotLib Compose design system.

Rocket: Mobile Performance Toolkit | Kotlin, Swift, Macrobenchmark, Perfetto

Vel Labs (internal)

Built a tool that benchmarks Android and iOS builds on real devices, reporting 54 Android and 12 iOS KPIs across 6 journeys against a baseline, driven by a YAML flow DSL.
Its analysis grew Baseline Profile coverage by 48.8% and showed notification launches took 2.1 s, not 0.7 s.

Languages: Kotlin, Java, Swift, C#, Dart, C++
Frameworks: Kotlin Multiplatform (KMP), Compose Multiplatform, Jetpack Compose, Decompose, Koin, Ktor, Coroutines/Flow, Unity as a Library (UaaL), Flutter, MVVM, Clean Architecture, Room, Retrofit
Tools: Macrobenchmark, Baseline Profiles, Perfetto, Sentry, Firebase, GitHub Actions, detekt, Gradle, Xcode, Postman, Figma

Smart India Hackathon 2023 Winner: Won for problem statement 1286 — incentivization of legal service providers and client onboarding platform.
HackerzStreet 1.0 — IEEE, Manipal University Jaipur: Secured 1st Runner-Up in a national hackathon. Built Android and web-based e-commerce solutions for local vendors.

Dayananda Sagar College of Engineering, Bangalore

Karnataka, India

Bachelor of Engineering in Electronics and Communication Engineering

2020 – 2024