Description:
MK AI Lab is an applied AI lab built inside a multi-hundred-million-dollar business operating across 20+ countries, now spinning out as an independent company. Unlike most AI startups, we don't start with a demo and go looking for a customer. We start with production systems already running inside a real global supply chain, and a committed anchor customer from day one.
Our systems in production today (details shared with shortlisted candidates under NDA):
- An autonomous freight bidding and pricing engine, live in daily operations
- A multi-tier, multi-piece rate calculation engine for complex trade-lane pricing
- A customs-compliance agent with documented operational value in a major market
Our mission: bring agentic AI to global logistics — an industry where every automated decision moves real freight and real money.
The Role
You will be our founding ML engineer — the engineering hire of the lab. You'll work directly with the founder to harden, extend, and scale the agentic systems already in production, and to build the next generation of them for customers beyond our anchor. This is a role for someone who wants ownership of real systems with real consequences, not a research sandbox.
What you'll do
- Own the core agentic architecture: multi-step agent workflows for quoting, bidding, pricing, and compliance
- Iterate our autonomous bidding/pricing logic toward measurable operational impact — better decisions, tighter reasoning traces, stronger evaluation
- Design and run evaluation pipelines for agent behavior: offline evals, regression suites, and production monitoring for long-running agent workflows
- Build and maintain retrieval and knowledge infrastructure (knowledge graphs, vector retrieval) over messy, real-world logistics data: rate sheets, customs schedules, carrier contracts
- Ship to production continuously — you'll deploy, observe, and fix systems that operations teams depend on daily
- Help shape engineering culture, tooling, and hiring as the team grows
Our stack
Python / FastAPI · LangGraph · Claude (Anthropic API) · MCP servers · Neo4j · pgvector · Supabase · DigitalOcean · local inference (Ollama) for sensitive workloads
What we're looking for
- 4+ years of software/ML engineering experience, with at least 1–2 years building LLM-powered systems that shipped to production
- Hands-on experience with agent frameworks (LangGraph, or equivalent orchestration you can defend), tool use / MCP, and structured retrieval (vector and/or graph)
- Strong Python and production engineering fundamentals: APIs, observability, testing, deployment — you've been on call for something that mattered
- Pragmatic evaluation mindset: you know how to tell whether an agent is actually good, not just impressive in a demo
- Comfort with ambiguity and ownership — you'll often be the only engineer in the room
- Bonus: experience in logistics, supply chain, fintech pricing, marketplaces, or other domains where automated decisions carry direct financial consequences
- Bonus: experience with knowledge graphs (Neo4j), self-hosted/local inference, or compliance-heavy domains
Credentials open the door; evidence of what you've shipped is what gets you the role. Exceptional builders from any background are still encouraged to apply.
What we offer
- Founding-team equity with standard 4-year vesting (1-year cliff)
- Competitive salary benchmarked to top-tier Pakistan startup market, paid in USD if preferred (final range set with the successful candidate)
- Production systems and a paying anchor customer from day one — your work goes live immediately in a large-scale global logistics operation spanning 20+ countries
- Direct partnership with the founder; no layers, no committees
- The rare early-stage setup: startup ownership and speed, with enterprise-grade problems and revenue already in place