Hiring managers
See measurable AI impact (hours saved, engineers enabled, awards won) backed by 10+ years of shipping production software.
AI Engineer · GenAI, MCP & Copilot · Full-Stack
I help engineering teams adopt AI tools, and build the ones that save them real hours.
I am Pankaj Dixit, an AI engineer with a full-stack foundation. I roll out, troubleshoot, and teach GenAI tooling (Copilot, MCP servers, LLM automation) and build it as production software, currently on the Microsoft Xbox account at Tech Mahindra, on an 10+ year engineering base. Remote-first, open to hybrid or on-site.
2 wks → 30 min
developer onboarding time, automated with an AI-powered MCP server
80+ hrs
engineering time saved per new developer
17+
engineers enabled to AI White Belt certification as AI Champion
Top 6 / 70+
CodeRush 2026 Finalist for IncidentIQ, an AI incident-intelligence platform
Positioning
AI impact, backed by real engineering.
This page moves from proof to projects, then into focus areas and direct contact. It answers one question for a hiring team fast: can Pankaj ship GenAI tooling that delivers measurable results, not just demos?
GenAI & MCPMeasurable impactProduction-ready
AI & platform teams
Understand how I turn GenAI and developer-experience ideas into deployed tooling that the whole team actually adopts.
Recruiters & LinkedIn visitors
Get a fast snapshot of AI work and recognition, then move into projects, the MCP case study, and direct contact.
Selected Work
AI tooling and full-stack delivery with measurable impact.
Every project below is framed by problem, contribution, and measurable result. The AI & Automation work shows what I build today; the full-stack projects show the production engineering depth behind it. Confidential client work is shown without direct access.
4 project stories in featured view.
Microsoft / Xbox · Tech Mahindra
Confidential Scope
AI-Powered Developer Onboarding MCP Server
New developers spent roughly two weeks on environment setup (VPN, Docker, authentication, builds, and linting) before their first real contribution.
Key Contributions
- Built a production MCP (Model Context Protocol) server in TypeScript with 21 tools that a new developer triggers from the Copilot chat in VS Code, simply by asking to complete their onboarding and project setup
- The server checks credentials and access/entitlement requests, detects what is missing (repo clone, software installs, configuration), and performs the setup itself
- Completes repository setup and wires in run tools, turning days of manual setup and doc-reading into a guided, self-serve flow reused for every new joiner
2 wks → 30 min
Onboarding time
80+ hrs
Saved per developer
Microsoft / Xbox · Tech Mahindra · CodeRush 2026 Finalist
IncidentIQ: Agentic AI for Incident Response
Incident management was roughly 80% manual: slow hand-filed tickets, about 30% of incidents misrouted, and 20 to 40 minutes of log reading before triage even began.
Key Contributions
- Architected a five-stage agentic pipeline (ingest, analyze, decide, act, learn) on the Microsoft Agent Framework and Azure OpenAI that takes a raw alert to a draft, merge-ready pull request
- Grounded every recommendation in retrieval over past incidents and runbooks (Azure AI Search), with a confidence gate so the agent escalates instead of hallucinating a fix
- Made human review mandatory before any merge, with a PII and secret scrubber and a full audit log on every action
<10 min
Alert to draft fix (POC)
95%
Routing accuracy (POC)
My own concept, architecture, and build. Full implementation details shared on request.
Microsoft / Xbox · Tech Mahindra
ShieldAI: AI-Augmented Pre-Release Safety Gate
Between 'staging tests pass' and 'approved for production', nothing verified the deployed app actually worked. Visual regressions, silent dependency failures, and 403/401 API errors slipped through green test suites.
Key Contributions
- Designed a CD pipeline gate that runs in parallel to the existing Playwright tests, adding about 30 seconds and zero time to the critical path
- Built it on a clear contract: deterministic checks (render, API health, TLS) block the deploy, AI checks (visual diff, failure diagnosis, risk score) only advise, and humans decide
- Shipped a dual-provider design (GitHub Models by default, Azure OpenAI for single-tenant) behind one config flag, with a one-line rollback if the gate ever misbehaves
30 sec
Gate runtime
None
Added release wait
My own concept, architecture, and build. Full implementation details shared on request.
Microsoft / Xbox · Tech Mahindra
Confidential Scope
Legacy UI Modernization (Angular to React)
An end-of-life Angular 1.x configuration portal (20+ admin pages) had drifted from the team's design system, forced full-page navigation for every edit, and was the last holdout in an otherwise React monorepo.
Key Contributions
- Leading the migration to React 18 and TypeScript (strict mode) with Redux and redux-saga, embedded in the team's existing codebase with zero backend changes
- Established reusable patterns (saga-per-domain, an LCE loading/content/error state envelope, inline edit modals, themed CSS variables) that other engineers now follow page by page
- Improved UX alongside the rewrite: dark theme, inline modals instead of route sprawl, distinct empty and error states, and accessibility on icon-only controls
20+
Admin pages
Zero
Backend changes
Ongoing migration. Feature parity reached across migrated pages; further details available on request.
Focus Areas
Where I create the most value.
For AI, platform, and product teams that need an engineer who can build GenAI tooling and ship it as dependable production software, then get the rest of the team using it.
GenAI Application & Tooling
LLM-powered apps, MCP (Model Context Protocol) servers, and Copilot-driven automation that remove manual steps from real engineering workflows.
AI Tools Support & Enablement
Copilot rollout, configuration, and troubleshooting, plus prompt-engineering coaching and AI Champion programs that raise an organization's AI readiness and certification rates.
Full-Stack Product Delivery
Production React/Next.js and TypeScript front ends, API integration, and cloud deployment on Azure, the engineering base that ships AI features for real.
How I Work
Start from the real workflow, not the demo.
1
Find the manual, repetitive workflow that is quietly costing the team hours.
2
Design an AI-assisted solution (MCP server, Copilot flow, or LLM automation) around the real task.
3
Build it as production software with guardrails, not a one-off script.
4
Measure the saved time and adoption, then enable the wider team to use it.
Open to Opportunities
Open to AI tools support, AI engineer, and GenAI-focused full-stack roles.
Reach out directly through email, LinkedIn, or GitHub. Recruiters and hiring teams can also send role details through the quick message form.
Remote-first · Available at short notice · Open to hybrid/on-site