I design and build complex infrastructure at the intersection of AI, data, and compliance — data platforms, automation pipelines, RAG systems, and the governance layers (RBAC/ABAC, encryption, audit trails, access control) that let enterprises actually trust and ship AI in production.
I'm not just shipping AI features — I'm building the plumbing underneath them: data ingestion, transformation, orchestration, and the compliance layer that makes AI safe to run in regulated industries like legal and healthcare.
At 10Pearls I own architecture for a compliance-grade legal case management platform. Through HKDev.co, my own agency, I deliver AI-native, compliance-grade systems for law firms, healthcare, and e-commerce clients across the US, UK, and UAE.
| Area | What it means in practice |
|---|---|
| Data Platforms & Pipelines | ETL, data transformation, and prep for ML training — turning messy enterprise data into something models can use |
| Agentic Automation | Multi-agent workflows and task-specific agents (LangGraph) that act on trained models, not just chat with them |
| Compliance & Data Governance | HIPAA/GDPR-compliant architecture, RBAC/ABAC, encryption at rest, audit trails, multi-tenant firm-layer isolation |
| Ontology / Semantic Layer | A semantic layer on top of raw/transformed data that defines entities, relationships, and business meaning explicitly — so agents and LLMs reason over structured knowledge instead of guessing from raw rows and tables |
| Digital Twin of the Business | A live virtual model of a company's operations, fed by the same data pipelines, that can be simulated, monitored, and queried by agents — moving from "AI that answers questions" to "AI that models how the business actually runs" |
| Full-Stack Product Engineering | Production systems end-to-end — Next.js/React frontends, Node.js/FastAPI backends, GraphQL, WebSockets |
- Enterprise AI Data Platform — lets enterprises bring their own data, transform and pipeline it for ML, and run task-specific agents on the trained models (e.g. an agent flagging underperforming wells in an oil refinery). Built on
Node.jsGraphQLBullMQInngest— data ingestion → transformation → training → agent orchestration, end to end. - Legal Case Management Platform (10Pearls, NDA client) — HIPAA-compliant, multi-tenant, with RAG document search, ETL pipelines, Keycloak RBAC/ABAC, microservices, and Temporal-orchestrated agentic workflows. I own the architecture, data modeling, firm-layer resolution, and encryption at rest.
Languages & Frameworks
Data & Infra
AI, Agents & Data
Compliance & Governance
| Project | What it is | Stack |
|---|---|---|
| Newstrick | Autonomous AI publishing pipeline | TypeScript, LLM orchestration |
| SentShare | Zero-knowledge encrypted file sharing | Node.js, cryptography |
| Playhouse | Multi-agent social platform | LangGraph, multi-agent systems |
| Splits59 | client project — e-commerce platform | Remix.js · Next.js · Shopify· Liquide . Graph APIs |
| Promotravle | client project Travel Toure Booking Platform | *Next js , Node js , Sanity , Sentry * |
| BusinessBuilder AI | HKDev.co client project | LangGraph • AI Agents • Prisma ORM • E2B • Vercel • Gemeni AI |
| AI Client Evaluation Platform | HKDev.co client project for a law firm — client evaluation/intake | Python • Fast APIs • Canvas UI • Browser Agents • Claude • Lang Chai |
(swap # for real repo/case-study links, and fill in the blanks — I don't have the specifics on the newer ones)
I write on AI/dev topics on dev.to, including a piece on proprietary data as a legal AI moat, and produce Roman Urdu/Hindi educational tech content for social media.