Full-stack engineer and AI agent builder. CEO at Contrive Solutions.
I build systems that file taxes, make phone calls, and open their own pull requests.

I run Contrive Solutions, an AI-first development agency with 50-plus engineers and offices in Lahore and Danville, California. I still ship code every week. That mix is deliberate. Sixteen years of building production systems taught me that architecture decisions made from a distance are usually wrong.

My work sits in two worlds. One is classic full-stack engineering at scale, things like a German tax platform running ten sharded PostgreSQL instances, or a music-industry asset pipeline that cut upload times by 90 to 97 percent. The other is agentic AI, where the software does the work instead of assisting with it. A recruitment agent I built took CV screening from 3-5 days down to under 4 hours and saved the client $140K a year. The voice sales agent did even better at $210K.

Most of my attention goes to autonomous engineering pipelines. I built a system where Claude Code picks up Jira tickets on a 600,000-line Laravel and Vue codebase, pulls context through custom MCP servers for Bitbucket, Jira, Slack, and Sentry, retrieves the right code via Qdrant vector search over tree-sitter AST chunks, then writes the fix and opens the PR. It currently closes about 40 percent of tickets with no human at the keyboard. The other 60 percent is why I still have a job.

flowchart LR
    A[Jira ticket] --> B[Claude Code agent]
    B <-->|MCP servers| C[Bitbucket · Jira · Slack · Sentry]
    B <-->|Qdrant vector search| D[tree-sitter AST chunks<br/>600K-line Laravel/Vue codebase]
    B --> E[Implementation + tests]
    E --> F[Pull request opened]
    F --> G{CI + human review}
    G -->|~40% merge clean| H[Shipped]
    G -->|Sentry errors feed back| B

I design AI features as operational software, with the same engineering discipline expected from any production system:

• Agent orchestration: deterministic state machines, tool calling, retries, checkpoints, approval gates, idempotency, and human-in-the-loop controls.
• RAG and search: ingestion pipelines, structural chunking, embeddings, hybrid retrieval, reranking, metadata filtering, citation grounding, and evaluation.
• Realtime AI: streaming, WebSockets, interruption handling, conversation state, function calling, latency optimization, and fallback paths.
• LLM reliability: structured outputs, schema validation, prompt/version management, caching, token budgets, rate-limit handling, model routing, and observability.
• Enterprise integration: REST, GraphQL, webhooks, event-driven workflows, queues, CRM/ERP integrations, payments, identity, RBAC, and audit trails.
• Cloud delivery: Docker, Kubernetes, CI/CD, infrastructure automation, monitoring, autoscaling, backups, incident debugging, and production hardening.

AI & Agents

Backend

Frontend

Data & Infrastructure

Some of the least glamorous work I do is the most valuable. ELSTER and DATEV integrations for German tax filing. KYC and AML flows for compliance platforms. SEPA payments. HIPAA-grade telehealth with BAA coverage and AES-256 at rest. Multi-tenant SaaS with sharded databases. If your project involves a regulator, I have probably argued with that regulator's API documentation before.

Fair warning, most of my client work lives in private Bitbucket repos that clients own, so these graphs undercount things badly. The interesting code is under NDA.

Contrive shows up in person. GITEX Global 2023 in Dubai, Web Summit Qatar 2025 in Doha, GITEX Europe 2025 in Berlin. I have also spoken at Pakistan Tech Talk about applied AI. Between events I run Contrive's 50-plus person team, and an unreasonable number of Claude Code sessions.

I am most useful when a product requires a combination of architecture, hands-on implementation, AI integration, cloud delivery, and engineering leadership.

Typical engagements include:

• adding production AI capabilities to an existing SaaS platform;
• designing RAG, semantic search, AI-agent, or conversational-data systems;
• modernizing legacy Laravel, Node.js, Python, React, Vue, or Rails applications;
• building multi-tenant SaaS, workflow automation, integrations, and real-time features;
• reviewing architecture, reliability, performance, security, and cloud cost.