{ "name": "Idrees Kamal", "location": "Chicago, Illinois, United States", "focus": [ "Applied AI", "AI operations", "AI enablement", "Workflow automation" ], "currentRole": { "title": "AI & Automation Engineer", "employer": "JDA TSG", "start": "2026-03", "scope": "Cross-functional AI strategy, integration delivery, and adoption infrastructure" }, "education": "Northwestern University, B.S. Computer Science, 2020", "contact": { "email": "ikamal97@gmail.com", "linkedin": "https://www.linkedin.com/in/idreeskamal" }, "caseStudies": [ { "id": "hr-integration", "category": "JDA TSG / ENTERPRISE OPERATIONS", "status": "Production integration", "title": "Reliable employee data. Controlled changes.", "short": "An HR lifecycle integration built around identity, exceptions, and verification.", "problem": "Employee lifecycle changes needed to move reliably between HR and performance systems. File-based handoffs and mismatched identities created correction work and operational risk.", "role": "I led the integration work: lifecycle routing, identity matching, source write-backs, exception handling, test design, controlled correction waves, and destination verification.", "decisions": [ "Treat create, update, deactivate, and reactivate as distinct paths with explicit matching rules.", "Test a small canary before expanding a change; define rollback and exception ownership before execution.", "Read the destination API after a change. A successful request alone is not proof of the intended record state." ], "result": "Delivered the core production sync and verified corrections across hundreds of employee records. Correction cohorts overlap and are not added together.", "limits": "This describes delivered capability and verified corrections, not quantified labor savings or a claim that every reconciliation exception was closed.", "stack": [ "Make", "REST APIs", "Lifecycle routing", "Canary testing", "Reconciliation" ], "proof": "Production workflow, controlled test results, and destination readbacks. Internal artifacts are confidential; architecture is summarized here.", "next": "Measure correction frequency, exception aging, and handling time against a pre-change baseline." }, { "id": "ai-adoption", "category": "JDA TSG / AI ENABLEMENT", "status": "Adoption infrastructure launched", "title": "Give people a way to use AI well.", "short": "A company-wide adoption program with practical learning, reusable prompts, and accountable rollout.", "problem": "Giving teams access to AI tools does not establish safe use or adoption. People need relevant examples, a place to learn, and support inside their existing workflow.", "role": "I developed the learning hub, prompt assets, role-based learning paths, community operations, and launch communications. I also designed measurement and rollout criteria.", "decisions": [ "Start with a controlled first wave and clear guidance on which tool to use.", "Build the learning and support surface in SharePoint and Teams, where employees already work.", "Separate delivered training infrastructure from actual usage, quality improvement, and time saved." ], "result": "Launched the learning hub and community infrastructure for an intended company-wide program of approximately 300 employees.", "limits": "The population is the intended program scope, not a count of active users or completed training. Adoption rates and business impact have not been established in this case study.", "stack": [ "Microsoft 365 Copilot", "SharePoint", "Teams", "Change management", "Measurement design" ], "proof": "Live learning surface, reusable prompt library, community channels, and launch communications.", "next": "Track active usage, training completion, repeat use cases, and task quality before scaling." }, { "id": "lookbook", "category": "KAVALIER / CUSTOMER EXPERIENCE", "status": "Used in the sales workflow", "title": "Help clients see what they are buying.", "short": "A personalized AI lookbook that connects a fabric conversation to a visual buying experience.", "problem": "Fabric swatches do not help every prospect imagine a finished custom garment on themselves. That uncertainty makes a consultative sale harder.", "role": "I selected the sales problem, designed the customer workflow, and built the application with AI coding tools. I owned integration, debugging, and verification.", "decisions": [ "Use a style profile and client photo to make the output personally relevant.", "Connect a React and TypeScript interface to an Express backend and the Gemini API.", "Treat generated imagery as a sales illustration; it is not a verified pattern, fit prediction, or manufacturing specification." ], "result": "The lookbook became a primary tool in the clothing sales process. Clients could discuss concrete visual options instead of relying only on swatches.", "limits": "The evidence supports real workflow use. No controlled conversion lift or exact revenue attribution is claimed.", "stack": [ "React", "TypeScript", "Express", "Gemini API", "Playwright" ], "proof": "Working application and an end-to-end test suite. A walkthrough can cover architecture, generation flow, and failure handling.", "next": "Measure time to a usable lookbook, generation failure rate, and consultation-to-order conversion." }, { "id": "launch-readiness", "category": "INDEPENDENT PROJECT / APPLIED AI", "status": "Reference prototype", "title": "Surface the blocker before the handoff.", "short": "A restaurant launch-readiness prototype that keeps findings tied to source evidence.", "problem": "A handoff can look complete while source documents disagree, approvals are missing, or evidence belongs to another location.", "role": "I selected the problem, built the prototype with Codex, defined structured output and citation checks, and created deterministic safeguards and reference scenarios.", "decisions": [ "Represent readiness as individual checks with supporting quotes, not an unqualified generated summary.", "Downgrade a finding when its citation does not match the supplied source text.", "Use fictional restaurant data and require human review before a launch decision." ], "result": "Built a reference walkthrough with six readiness checks, conflict and missing-evidence scenarios, structured output, and six passing deterministic safeguards in the September 4 review.", "limits": "This is an independent prototype, not an Owner product or a production deployment. Live model evaluation was blocked by API credits in the last recorded test; reference examples are hand-authored, not model outputs.", "stack": [ "Next.js", "OpenAI Responses API", "Structured output", "Citation validation", "Evaluation harness" ], "proof": "Explore the synthetic reference example below. It demonstrates the intended review workflow and does not call an AI model.", "next": "Run the blocked model cases, inspect false-ready decisions, and evaluate in shadow mode before any operational pilot." }, { "id": "vicegerent", "category": "VICEGERENT / BUSINESS OPERATIONS", "status": "Delivered operating systems", "title": "Build the systems. Run the business.", "short": "Customer records, booking, and digital operations grounded in firsthand operating responsibility.", "problem": "Customer information and scheduling were fragmented across spreadsheets and manual processes in a traditional custom-clothing business.", "role": "I migrated customer records, connected booking to the CRM, rebuilt the website, and ran sales and daily operations solo for substantial parts of the year.", "decisions": [ "Use a practical Notion CRM with standardized customer fields and lifecycle stages.", "Connect Calendly bookings to the CRM through Zapier so customer context follows the appointment.", "Design workflows around real consultations, fittings, orders, and follow-through." ], "result": "Migrated more than 750 client files. Recorded solo sales were approximately $57,600 across 36 clients during the 2022\u20132023 reporting period. Website engaged sessions increased from 1,878 to 3,281 in the recorded comparison.", "limits": "Solo revenue reflects personally handled business, not incremental revenue attributed to automation. The web comparison does not establish a causal experiment.", "stack": [ "Notion", "Zapier", "Calendly", "WordPress", "Google Analytics" ], "proof": "Historical operational and analytics records inform this summary. Private client records are not published.", "next": "Use matched measurement periods and distinguish personally handled revenue from incremental lift." } ], "limitations": [ "Production LLM and model-evaluation depth should be assessed directly.", "SQL is foundational.", "People-management experience includes supervising three junior business analysts; no large engineering organization management is claimed." ], "updated": "2026-09-08" }