Arjun Sabharwal

Skoda Volkswagen India — IBM Platform (Project Ruby)

Client: Skoda Volkswagen India Pvt Ltd  |  Employer: Lauren Information Technology

Project Objective

Implement three IBM enterprise platforms — Planning Analytics, RPA (Robotic Process Automation), and Datacap — within Skoda Volkswagen India's operations, while introducing a structured Agile delivery model for the first time across all workstreams.

My Role & Responsibilities

• Managed infra setup, requirements gathering, SIT, UAT lead, and final sign-off across all phases
• Led code reviews, architectural reviews, testing/QA oversight, and stakeholder demos
• Introduced transparent Agile processes — breaking large stories into sprint-sized tasks, guarding the team from unnecessary meetings, and splitting the team into three independent sub-teams to reduce bottlenecks
• Championed CI/CD pipeline for RPA enabling per-sprint releases and module-level sign-off
• Spearheaded local datacenter environment setup and managed ambiguity buffer for new technology onboarding
• Scoped and billed all change requests (CR) with revised commercials to protect project timeline
• Ran bi-weekly stakeholder demos and steering committee meetings for client alignment

Key Outcomes

• Sprint velocity increased by 25% through Agile process transparency and single-task focus
• Cycle time reduced by 15%
• On-time delivery of three major platform releases

StackIBMAWSJira

IndusInd Bank — Online Approval System

Client: IndusInd Bank  |  Employer: Lauren Information Technology

Project Objective

Design and deliver a digital approval workflow system for internal banking memos and invoices, replacing a manual process with a structured, multi-tier sequential approval engine integrated with the bank's Document Management System (DMS).

My Role & Responsibilities

• Attended and led requirement gathering sessions, producing the BRD, FRS, and SRS documents
• Created and maintained the Risk Register, Opportunity Register, Change Request Log, and a PMI-aligned Project Management Workbook covering milestones, phases, and timelines
• Ran all Scrum ceremonies (sprint planning, daily standups, retrospectives, reviews) for a 10-person team: 2 FE · 3 BE · 1 SA · 1 QA · 1 BA · 1 DevOps
• Delivered SteerCo presentations to senior bank stakeholders on progress and risks
• Monitored day-to-day delivery and ensured quality across front-end, back-end, and QA tracks

Key Outcomes

• Delivered a fully functional sequential approval portal with user dashboard, case creation, multi-approver flow, and DMS document attachment
• All PMI project management artefacts maintained throughout the lifecycle

StackReactJSJavaSpring BootMicroservicesJira

AskTara — Alembic WhatsApp LLM Chatbot

Client: Alembic Pharmaceuticals  |  Employer: Lauren Information Technology

Project Objective

Build a WhatsApp-based conversational AI system enabling Alembic's healthcare operators to retrieve information from a large corpus of PDF documents and structured databases, using a multi-LLM architecture with guardrails.

System Architecture

• Layer 1 — Language Conversion (GPT-4o Mini): Any language input is converted to English; the model also routes queries to the appropriate downstream engine
• Layer 2 — English → SQL (LLM): Structured database queries generated by LLM, executed against the database, and results returned
• Layer 3 — Results → Natural Language (LLM): Table-based results converted into clear, human-readable English responses
• Document Retrieval: Amazon Kendra used for semantic search across PDFs
• Guardrails: AWS Bedrock guardrails personally configured and tested

My Role & Responsibilities

• Led prompt engineering, technical scoping, and system architecture
• Managed all PMI project management activities — daily syncs with customer and internal team, Agile ceremonies, scope documentation
• Curated stakeholder demos and produced status reporting decks
• Personally handled Bedrock guardrail configuration, testing, and validation

StackOpenAIAmazon KendraAWS BedrockWhatsAppNode.jsPython

Autodesk — Digital-HIG · Screencast · Assistant

Client: Autodesk Inc  |  Employer: Carnera Technologies

1. Digital-HIG — Autodesk Design System

Company-wide React component library used by all Autodesk engineering teams to build any UI product.

• Led a Scrum team of 7; ran all sprint ceremonies and bi-weekly stakeholder calls with Directors, Sr. Managers, and PMs (200+ attendees in monthly progress presentations)
• Managed a support channel handling 437 requests/year — 90% resolved within 1 day (tech support) or 1 week (dev work)
• Launched 80 new UI components; upgraded codebase from MUIv4 → MUIv5
• Reviewed every major PR (100+ lines changed) and all new component final PRs, leaving detailed code quality comments
• Raised code coverage from 75% → 90%; reduced sprint spillovers by 18%; improved deployment time by 25%
• Vetted and onboarded new developers with a structured interview and scoring system
• Produced 18 tutorial videos (recorded & edited) for complex components; collaborated with a technical writer on full documentation
• Achieved 760+ internal npm installs

StackReactJSStorybookJestJenkinsGitHub ActionsD3.jsJiraFigma

2. Autodesk Screencast — App Modernisation

E-learning platform for recording and sharing mouse/keyboard walkthroughs of Autodesk software (part of the AKN initiative). Legacy codebase from 2014.

• Led end-of-life modernisation with minimal documentation; broke down legacy code into logical, actionable items for a team of 4
• Identified deprecated library risks and collaborated with cross-functional architects on risk assessment and replacement strategy
• Delivered 12 new APIs aligned with Autodesk rebranding; completed UI redesign balancing design vision with timeline constraints
• Improved application load time by 175% via modern libraries and Webpack minification
• Patched 63 security bugs from deprecated code; upgraded Java to the latest version resolving 30 bugs (12 breaking)
• Documented all APIs and front-end tools; moved project to LTS (retirement mid-2023)

StackjQueryJavaJenkinsGitHubJira

3. Autodesk Assistant — AI Help Tool

Autodesk's built-in AI assistant embedded across the main site and consumer-facing pages for purchase decisions, sales routing, and customer service.

• Led code coverage improvement initiative: 43% → 90% across 150 React components broken down component-by-component into sprint tasks
• Completed security clearance tasks with 5 developers over 3 months
• Reviewed 39 backend APIs for performance bottlenecks; guided break-up of large component files (max 125 lines policy)
• Reviewed all 40 PRs with component changes; consulted architect on questionable architectural decisions
• Advocated for and implemented a pipeline code coverage gate — no code merged without minimum 90% coverage
• Eliminated all random crashes post code changes; ran multiple linters for sustained code quality

StackReactJSNode.jsPythonJiraConfluenceGit

Pixly.ai — Construction Photo Documentation App

Client: Pixly (CEO: Lee Mills)  |  Employer: Carnera Technologies

Project Objective

End-to-end product management and Go-To-Market delivery for Pixly.ai — an app allowing construction site teams to catalogue sites via images, organise them into sets, and share/tag/sort them. Delivered across Android (React Native), iOS, and Angular web app.

My Role & Responsibilities

• Handled full Product Manager role alongside Go-To-Market strategy for first corporate clients
• Met with CEO Lee Mills twice a week, capturing pain points and building feature roadmaps at least 1 month in advance
• Balanced feature development with bug fixes across sprints; maintained weekly feedback integration cycle
• Ran all Scrum ceremonies for a team of 3 (Java · iOS/Frontend · React Native/Android)
• Bi-weekly stakeholder status meetings with CEO and CTO of Carnera
• Delivered two full development phases; kicked off third phase dedicated to Android
• Created migration plan to retire iOS app and move to a hybrid solution for both mobile platforms

Key Outcomes

• 520% customer gain during our management (first corporate GTM phase)
• 70 new corporate customers and 220 new individual customers
• Won 2 additional contracts off the back of the first project due to efficient delivery

Stack React Native Angular Java

TheLink.ai — AI/ML on Construction Spec Sections

Client: TheLink.ai (CEO: Hugh Seaton)  |  Employer: Carnera Technologies

Project Objective

AI-driven platform for processing construction spec sections (a document type specific to US construction companies). Uses AI/ML to identify and extract spec sections from thousands of pages of documents, outputting results in a tabulated, searchable list — reducing hours of manual review to a single click.

My Role & Responsibilities

• Managed a team of 5 (2 React · 2 Java · 1 QA) using Agile/Scrum
• Weekly meetings with CEO Hugh Seaton — triaging requirements and UI changes alongside backend tasks
• Deep domain research into construction spec documents; worked with a test dataset of 200 documents × 100 pages each
• Acquired strong understanding of construction industry-specific tooling and workflows
• Wrote documentation required for publishing the application to its respective app platform

Key Outcomes

• Launched 12 features into the application
• Full documentation produced for platform publishing

Stack Python LLM / AI/ML ReactJS Jenkins Jira Confluence