Padmil • AI PM• •Enthusiast •
Technical Product Leader · AI/ML · San Francisco, CA

Padmil Khandelwal — AI Product Manager, Agentic AI & Full-stack Developer

Padmil Khandelwal

Full-stack Engineer and Data Scientist turned Enterprise AI Product Leader. 7+ years building and scaling B2B SaaS — delivering $25M+ in platform impact, 30%+ efficiency gains, and 500K+ weekly transactions at Oracle.

Scroll to explore

About

/ Summary & Highlights

Full-stack Engineer and Data Scientist turned Enterprise AI Product Leader. 7+ years building and scaling B2B SaaS — delivering $25M+ in platform impact, 30%+ operational efficiency gains, and 500K+ weekly transactions across agentic AI, conversational AI, and payments products at Oracle.

7+
Years Experience
500K+
Weekly Transactions
$25M+
Platform Impact
30%+
Efficiency Gains

Experience

/ Career Journey

Oracle

Senior Product Manager, Fusion ERP Financials

Mar 2021 – Present

USA

  • Owned product strategy and execution for AI-driven Accounts Payable (Procure-to-Pay) workflows, reimagining invoice processing using multi-agent AI systems powered by event-driven orchestration across 500K+ weekly transactions, reducing manual overhead by ~$300K/month.
  • Partnered with the AI agent platform team to identify capability gaps and drive feature enhancements, including introducing human-in-the-loop (HITL) frameworks to ensure reliability and auditability in AI-driven financial workflows, enabling enterprise-grade deployment of agentic systems in high-risk financial operations with rapid adoption across 40+ agents.
  • Partnered with UX and research to create reusable design building blocks & workflow primitives for an agentic enterprise SaaS platform, enabling 15+ teams to design and launch AI-powered automation 6 months faster and standardizing workflows across ERP products.
  • Led development of an LLM-powered conversational agent to analyze $50M+ in outgoing payments data, simulate what-if scenarios for working capital optimization, and surface supplier-specific insights enabling finance teams to make proactive cash flow decisions within procure-to-pay workflows.
  • Owned enterprise search and data access platform (Elasticsearch) across 15+ business objects, improving retrieval success by 40% and reducing latency by 60%, enabling semantic and AI-powered search experiences for enterprise knowledge discovery.
  • Developed a scalable data ingestion and enrichment platform for unstructured documents (OCR and multimodal pipelines), improving data quality by 40% and enabling downstream transaction creation with minimal user input.
  • Defined and operationalized LLM evaluation frameworks (LLM-as-a-judge) for GenAI-powered expenses and payables agents, establishing AI governance through systematic measurement of model accuracy, faithfulness, and reasoning, and developing custom scoring functions that improved task completion rates by 15% across production workflows.
  • Led development of a B2B virtual card payments product in partnership with financial institutions and card networks, introducing a new supplier payment method that improved working capital by ~10% and drove $5M+ in initial transaction volume.
  • Built an extensible payments infrastructure supporting 5+ digital payment methods (credit cards, PIX, UPI, PayPal), enabling scalable transaction processing exceeding $15M/month and expanding global supplier payment capabilities.

IBM

Senior Data Scientist Intern, Global Business Services

June 2020 – Aug 2020

USA

  • Built an enterprise search platform using headless web crawlers (Puppeteer, Node.js) and OCR-based ingestion, replacing legacy Google Enterprise Search and scaling to 10K+ daily queries for internal knowledge discovery.
  • Developed and validated a predictive ML model (Random Forest) to identify high-risk customers (93% accuracy), performing feature engineering and model tuning to enable proactive engagement workflows that reduced uncollectible receivables by $500K+.
  • Designed a service orchestration engine integrating Twilio SMS and Watson Assistant to trigger AI-driven customer engagement, operationalizing closed-loop workflows from prediction to action.
  • Delivered a conversational AI solution to recommend optimal rate plans for customers based on churn risk and usage patterns, driving $1.2M+ in projected savings.

Larsen & Toubro Infotech Ltd

Software Engineer Full Stack, Fosfor

July 2017 – June 2019

India

  • Led a 5-member engineering team to deliver a self-service BI module, applying design thinking to empower non-technical users and reduce data analyst dependency by 25%.
  • Architected and built a cloud-native SaaS data analytics platform processing 5TB+ of enterprise data, enabling self-serve insights and improving time-to-insight by 40% for business users.
  • Built a flexible AI/ML platform with intuitive UI to streamline model training and deployment workflows, reducing time-to-production for data science applications by 4x.
  • Owned the product roadmap for a data wrangling and transformation feature, while leading full-stack development (Angular, Spark-based pipelines) to reduce data preparation time by ~30%.
  • Collaborated with UX and end-users to improve product usability, translating feedback into design enhancements that increased feature adoption by 80%.
  • Owned delivery as Scrum Master for a 5-member engineering team, defining user stories and prioritizing backlog while contributing to development, reducing bugs and backlog by 25%.

Education

MS in Information Systems Management · Carnegie Mellon University

USA

BS in Computer Science with Specialization in Analytics · University of Petroleum and Energy Studies

India

Labs

/ Selected projects & explorations

Claude-to-Codex Skill Converter — Developer Tools project by Padmil Khandelwal: Convert any Claude skill into a ready-to-use Codex skill — automatically. Search
Developer Tools

Claude-to-Codex Skill Converter

Convert any Claude skill into a ready-to-use Codex skill — automatically. Search by name, fetch, transform, validate, and install in one command.

Multi-Agent Invoice Orchestration — Agentic AI project by Padmil Khandelwal: Autonomous multi-agent platform resolving Invoice Exceptions across 500K+ weekly
Agentic AI

Multi-Agent Invoice Orchestration

Autonomous multi-agent platform resolving Invoice Exceptions across 500K+ weekly transactions via orchestration logic and dynamic tool-calling.

Internal
LLM Ingestion Engine — Infra & ML project by Padmil Khandelwal: High-scale multi-modal ingestion layer processing millions of unstructured files
Infra & ML

LLM Ingestion Engine

High-scale multi-modal ingestion layer processing millions of unstructured files. 40% OCR improvement to provide high-fidelity context for downstream AI agents.

Internal
RAG + LLM-as-a-Judge — GenAI project by Padmil Khandelwal: GenAI and RAG prototype with automated LLM benchmarking optimizing for faithfuln
GenAI

RAG + LLM-as-a-Judge

GenAI and RAG prototype with automated LLM benchmarking optimizing for faithfulness and reasoning accuracy. Boosted goal completion by 20%.

Internal

Let's
connect

/ Get in touch

Email

khandelwalpadmil@gmail.com

Location

San Francisco, CA