I build things,
ship them,
and sometimes sell them.
Senior AI Product Manager with 9+ years delivering enterprise AI across UAE, KSA & India. Co-founded Visiolab, acquired by Droom. Built The Learning Buddy from scratch with 2 people, scaled to 15,000+ users. Now leading AI transformation at EMB Global.
Featured Work
Things I built that matter
Exit Story
Built & Sold to an Indian Unicorn
Visiolab → Droom · 2019–2021
Co-founded Visiolab, building an AR automotive showroom platform using Unity + OpenCV. Ran paid POCs with 10 OEMs to validate demand, then secured a cash + stock acquisition by Droom within 12 months. Continued as EIR post-acquisition to integrate the platform into Droom's ecosystem.
Founder Story
Built an EdTech App from Scratch — 2 People
The Learning Buddy · 2022–2023
Built the iOS and Android app end-to-end in Unity before vibe coding existed. Led a team of 25 across Unity devs, 3D artists, UI designers, and marketing. Raised innovation funding from the Delhi Government. Reached #9 top education app in India on Google Play Store.
AI Deployment at Scale
100-Kiosk AI Ordering System in Saudi Arabia
Barns KSA — via EMB Global · 2023–2024
Built the full AI orchestration pipeline from scratch as PM: STT → turn detection → facial detection → fine-tuned Qwen LLM (Saudi dialect) with structured tool-calling → TTS output. Deployed across 100 kiosks, enabling 40% more orders per day through bilingual AI interaction.
Enterprise Delivery
Built a Multi-Million Auto Marketplace in the UAE
AD Ports Group · 2025–Current
Led AD Ports Group's automotive marketplace from presales to delivery. Ran a cross-border program with sellers in China, buyers in the UAE, and stakeholders across Turkey and the UAE. Owned product build from scratch, multiple partner integrations, Azure deployment on the client's environment, and delivery through a 10-person tech team.
Agentic Systems
I design AI that can take action
The work I enjoy most sits beyond chat. It involves multi-step reasoning, tool use, retrieval, state management, and production constraints like latency, reliability, and operational ownership.
Voice Agents
Multilingual Voice Orchestration
Production-grade voice systems that listen, reason, call tools, and respond in real time across noisy physical environments.
Agent loop
100 kiosks deployed in KSA
Enterprise Agents
Document and Workflow Automation
Agentic systems that combine retrieval, structured outputs, and human review gates to reduce manual operations in legal and enterprise workflows.
Agent loop
Multi-tenant architecture for business use
Operational Agents
Action-Taking Internal Copilots
Agents designed to complete work, not just answer questions: triage requests, coordinate tools, and trigger downstream business actions.
Agent loop
Designed for production workflows, not demos
Career
The full journey
Agentic Builds
Applied AI systems I built hands-on
These are not generic chatbot demos. Each project is framed as an agentic workflow with orchestration, tool usage, state, and architecture decisions that map to real production systems.
Case study
Realtime Voice Operations Agent
Needed a voice-native assistant that could handle live back-and-forth, preserve context, and trigger downstream actions instead of behaving like a simple speech bot.
Agent loop
Streams audio, interprets intent, calls tools, updates session state, and returns spoken responses fast enough for a natural conversation loop.
Architecture
Showcases low-latency orchestration and tool use in realtime interfaces.
Case study
Multi-Tenant Contract Review Agent
Legal workflows needed a system that could ingest tenant-specific documents, retrieve the right context, and flag risk with structured outputs instead of vague summaries.
Agent loop
Indexes documents per tenant, retrieves relevant clauses, runs guided review steps, produces redlines or risk flags, and escalates uncertain cases for human review.
Architecture
Demonstrates grounded reasoning, memory isolation, and controlled enterprise outputs.
Case study
Autonomous Email Coordination Agent
Scheduling threads waste time because intent, participants, timezones, and missing details are scattered across messy email conversations.
Agent loop
Parses inbound mail, extracts participants and constraints, plans the next action, coordinates with calendar tools, and sends confirmed invites without manual intervention.
Architecture
Shows end-to-end task completion with asynchronous tool calls and validation.
How I Build
How I build AI products that survive reality
My bias is toward systems that can be deployed, monitored, and defended in front of real stakeholders, not just good-looking prototypes.
01
Frame the job clearly
Start with the business action the agent must complete, the failure modes that matter, and the degree of autonomy that is actually acceptable.
02
Design the loop
Break the system into observe, reason, act, and verify stages. Decide what should be prompt-driven, tool-driven, or handled deterministically in code.
03
Ground with context
Use retrieval, structured memory, and system state so the model acts on current facts instead of improvising on stale context.
04
Control risk
Add validation, fallback states, human review gates, and output constraints so the agent remains usable under real production pressure.
05
Measure and iterate
Track latency, task completion, escalation rate, cost, and accuracy. Tune prompts, tool schemas, and workflow boundaries based on evidence.
Skills
What I bring to the table
Delivery
AI & Product
Stakeholder
Orchestration
Who I Am
I am at my best when the problem is real
Titles matter less to me than ownership. I tend to move toward difficult situations, whether that means aligning messy stakeholders, fixing broken delivery, or figuring out a path where none is obvious yet.
Personal Operating System
Problem solver by default
Personally and professionally, I do not run away from hard problems. I like stepping into ambiguity, breaking things down, and staying with the work until there is a path forward.
Built on discipline
Sport has shaped how I work: consistency, resilience, and the willingness to keep going when the easy option is to stop.
Curious outside work
I play guitar, enjoy long road trips, and generally look for experiences that slow things down enough to think clearly and come back sharper.
Outside Work
Those parts of my life feed the same trait that shows up in work: patience under pressure, stamina over time, and a bias toward momentum.