Aqueel Kanchwala

The problem

Enterprise process documentation requires translating business process knowledge into structured and maintainable process definitions.

What I did

• Built a Process Documentation Agent using the Skan ProcessDoc API.
• Integrated the agent with Skan process documentation capabilities.
• Worked with ProcessDoc and Process Definitions.
• Automated process documentation workflows.

Impact

• Developed an AI agent for automated process documentation.

The problem

Enterprise negotiation workflows require structured business information and integration with existing enterprise systems.

What I did

• Built an Autonomous Negotiator agent.
• Integrated the agent with enterprise systems.
• Developed agentic workflows for negotiation-related business processes.

Impact

• Developed an enterprise-integrated Autonomous Negotiator agent.

The problem

The organization needed to identify operational inefficiencies and opportunities for automation across its Procure-to-Pay processes.

What I did

• Configured Skan for P2P processes.
• Conducted process analysis and identified automation opportunities.
• Developed SQL/Python-based analysis.
• Used RPA Feasibility dashboards to support automation prioritization.

Impact

• $1.5M in realized / attributed savings.

The problem

Fragmented procurement information and disconnected enterprise systems limited supplier intelligence, negotiation effectiveness and data-driven procurement decisions.

What I did

• Designed an AI-native procurement transformation roadmap.
• Defined capabilities across supplier intelligence, negotiation automation, risk management and award optimization.
• Designed a multi-agent procurement architecture.
• Developed implementation roadmaps and business cases.
• Defined value realization frameworks and transformation KPIs.

Impact

• Architecture designed for 200K+ suppliers.
• $5B+ addressable spend.
• Targeted 0.5–1.5% sourcing savings.
• $40M–$75M annual value realization opportunity.
• Targeted 40–60% cycle-time reduction.

The problem

Supplier, sourcing, contract, performance and risk information was distributed across enterprise systems, limiting unified analytics and AI-enabled decision-making.

What I did

• Architected a Databricks-based procurement intelligence platform.
• Designed data integration across procurement domains.
• Worked on production implementation components.
• Defined the data foundation for supplier intelligence and downstream AI agents.

Impact

• Platform architecture designed to support 200K+ suppliers.
• Established a data foundation for a $5B+ procurement transformation program.

The problem

Procurement teams needed specialized intelligence capabilities to support supplier discovery, negotiations, market analysis, risk evaluation and sourcing decisions.

What I did

• Designed a multi-agent procurement architecture.
• Defined specialized agents for Supplier Discovery, Negotiation Intelligence, Market Intelligence, Risk Assessment and Award Optimization.
• Mapped agent capabilities to procurement workflows and enterprise data sources.
• Connected the architecture to the broader procurement intelligence platform.

Impact

• Defined an agentic architecture supporting enterprise procurement transformation across $5B+ addressable spend.

The problem

Procurement teams needed AI-assisted intelligence and contextual decision support for supplier negotiations.

What I did

• Designed an AI-powered procurement negotiation accelerator.
• Developed RAG-based solution components.
• Worked on LLM-powered negotiation workflows and supplier intelligence concepts.
• Contributed to the solution's innovation challenge submission.

Impact

• Achieved Top-3 placement in the EY Innovation Challenge.

The problem

AI workflows needed structured access to supplier management functionality within the HICX enterprise platform.

What I did

• Designed and implemented an HICX MCP Server.
• Exposed supplier management functionality through structured tools.
• Integrated enterprise supplier management capabilities into AI workflows.
• Took the integration through production deployment.

Impact

• Production deployment enabling AI workflows to interact with HICX supplier management capabilities.

The problem

Supplier Information Management implementations involved lengthy configuration and integration cycles, with repeated implementation effort across programs.

What I did

• Developed a reusable HICX SIM implementation accelerator.
• Integrated HICX and SAP ECC capabilities.
• Standardized implementation configurations and reusable components.
• Supported adoption across multiple client programs.

Impact

• Implementation timelines reduced from 24 months to 6 months across multiple client programs.

The problem

A global enterprise required consistent supplier onboarding, information management and governance across a large supplier ecosystem.

What I did

• Owned functional requirements and solution design.
• Defined HICX configuration strategy.
• Led testing and deployment activities.
• Supported post-go-live stabilization.
• Coordinated functional stakeholders and implementation workstreams.

Impact

• Functional delivery supporting a supplier ecosystem of 200K+ suppliers.

The problem

Insurance claims workflows involve multiple process steps and operational handoffs, creating opportunities for process discovery and efficiency improvements.

What I did

• Configured Skan for Insurance Claims processes.
• Worked with ProcessDoc and Process Definitions.
• Conducted process discovery and analysis.
• Identified bottlenecks and automation opportunities.

Impact

• Delivered process intelligence and automation assessment capabilities for Insurance Claims.

The problem

Patient intake workflows involve multiple operational steps, information capture and handoffs that can be assessed for process efficiency.

What I did

• Configured Skan for Patient Intake processes.
• Worked with ProcessDoc and Process Definitions.
• Conducted process analysis and process discovery.
• Identified workflow bottlenecks and potential automation opportunities.

Impact

• Delivered process intelligence and automation assessment capabilities for Patient Intake.

The problem

Organizations needed a structured way to identify processes suitable for automation and prioritize opportunities based on operational data.

What I did

• Developed RPA Feasibility dashboards.
• Conducted SQL/Python-based data analysis.
• Evaluated business process automation opportunities.
• Supported prioritization and value realization discussions.

Impact

• Enabled structured automation feasibility assessments across enterprise processes.

The problem

Enterprise procurement teams needed structured supplier performance capabilities and consistent approaches to supplier evaluation.

What I did

• Served as a functional SME for Supplier Performance Management.
• Worked on HICX initiatives, grids and functional configuration.
• Contributed to solution design and thought leadership.
• Supported internal SPM Center of Excellence activities.

Impact

• Developed reusable functional expertise and contributed to SPM solution development.

The problem

Large-scale supplier migrations and enterprise integrations required extensive data validation, reconciliation and reporting.

What I did

• Developed Python and SQL automation solutions.
• Automated supplier data validation and reconciliation.
• Supported data migration and reporting activities.
• Worked on large supplier datasets.

Impact

• Processed 1M+ supplier records.

The problem

Enterprise organizations evaluating procurement transformation needed solution demonstrations, technical discussions and business cases.

What I did

• Supported 10+ transformation opportunities.
• Conducted customer discovery and CXO workshops.
• Contributed to solution demonstrations and architecture discussions.
• Developed GTM decks, RFP responses and HICX solution materials.

Impact

• Supported 10+ enterprise transformation opportunities.

A personal portfolio designed so that nothing is hard-coded: the profile, skills and projects all live as structured content and are edited through a private admin portal.

How it works

• Static site built with Astro, with content stored as JSON and validated against typed schemas at build time.
• A Git-based admin (Sveltia CMS) behind GitHub login: each edit is saved as a commit, and Netlify rebuilds and publishes the site within about a minute.
• No database or backend to run or pay for.

Features

• Clickable skills and project tiles that open pop-ups with details
• Separate enterprise and personal project sections, with link fields only where the work is public
• Sections hide themselves when they have no content
• Mobile-first layout with automatic light/dark mode

Built with AI-assisted development using Claude Code.

Ops leaders track KPIs like claim turnaround and invoice-to-pay time, but finding out why one is slipping takes analysts days. Ask an LLM and you get a fluent answer with made-up numbers. WatchDog splits the work: the model forms hypotheses, weighs evidence and writes, and code does every calculation.

How it works

• Watch: scheduled KPI monitors check client policy thresholds. The most critical alert is investigated automatically, within a daily budget.
• Investigate: a LangGraph agent picks hypotheses from a fixed menu of checks. Code runs them, including confounder controls, and rule overrides catch the model's mistakes in either direction.
• Recommend: each action cites the policy clause behind it, and a deterministic what-if on the event log estimates its impact.
• Act: an action runs only after someone with the right business role approves it. Agents cannot approve, and the proposer cannot approve their own action. A follow-up check confirms whether the KPI moved.

Two processes, one engine

The engine is tested on two domain packs, each with 18,000 synthetic cases and planted root causes plus decoys: insurance claims (5-day SLA) and procure-to-pay (10-day invoice target).

Results

• 100% recall of the planted root causes and 0 decoys picked, on both packs
• 0 unverified numbers in the summaries
• 96–100% of policy answers cite a relevant passage
• Top-5 retrieval hit rate of 97–100% using hybrid search

Under the hood

Three MCP servers: Analytics and Knowledge are read-only, and Operations runs governed actions. It also has role-based access, a SHA-256 hash-chained audit log, sandboxed exploratory SQL labelled as unverified, observability for latency, tokens and quota, and 258 tests in CI.