AI Portfolio

Mahlab Maniar

Strategy and operations professional with 5+ years across management consulting (BCG), tech (Alibaba/Daraz), and Trella, a YC-backed marketplace. I took real problems I've worked on and rebuilt them as governed, judgment-aware AI systems.

Full deck ↓
The three pieces
Three problems I've actually worked on — each one a different call about where AI belongs in a real process.
01
Live

Carrier Onboarding & Qualification

Trella experience
Deterministic vendor checks, agent qualification, owner routing. A near-illegible WhatsApp submission becomes an auditable Tier 1 decision — no manual ops effort on the standard path.
carrier-onboarding.vercel.app →
02
Live

Deal & Pricing Governance Reviewer

Trella experience
Rate-card policy checks with AI triage on exceptions only. ~77% of the book clears with zero model calls; the rest is routed to the right owner with a written rationale.
deal-pricing-reviewer.vercel.app →
03
Live

Driver-Based Budgeting Tree

BCG experience
AI reads messy GL exports and rate-card PDFs; deterministic code does every calculation. A different domain, a different architecture — the set's breadth play.
camp-cost-model.vercel.app →
Full walkthrough
The complete portfolio deck
17 slides — problem, architecture, output, eval and honest limitations for all three pieces.
Open deck (PDF) →