Founding AI Systems Lead

Ark Abroad is building an AI-native operating system for a service business.

We're not looking for someone to build another chatbot.

We want someone who can take messy business problems, understand how the company actually operates, and turn them into reliable AI systems that remove work, improve decisions, and make a 20-person company operate with dramatically more leverage.

This is a builder + technical leader role.

You will still write code. You will prototype. You will debug. You will ship.

But you will also decide what should be built, how systems fit together, what should not be built, and eventually lead a small team around those systems.

What You'll Own

1. Make our CRM AI-native

Zoho is our operational source of truth.

Your job is to make AI capable of actually operating on top of it.

Examples:

• Agents that read CRM context before acting
• Automatically summarising calls and updating records
• Detecting risks, missing actions and operational failures
• Triggering workflows based on conversations and behaviour
• Surfacing the right information to sales/coaching/operations teams
• Allowing humans and agents to interact with the same underlying operational data

We don't want isolated AI tools.

We want AI embedded into how the company operates.

2. Build Ark's organisational brain

We have years of:

• founder knowledge
• coaching conversations
• sales calls
• SOPs
• research
• Loom videos
• ChatGPT/Claude conversations
• customer interactions
• outcomes and failures

We want to turn this into a governed intelligence layer that agents can reliably reason over.

You'll work on:

• ingestion pipelines
• canonical knowledge
• RAG/retrieval
• memory
• knowledge graphs where appropriate
• source attribution
• evaluation
• knowledge updating/deprecation
• context engineering

The objective isn't chat with our documents.

The objective is for organisational knowledge to become usable infrastructure for agents.

3. Build real agents

Examples include:

• learner research agents
• candidate positioning agents
• proof-of-work agents
• coaching preparation agents
• accountability agents
• sales intelligence agents
• pre-onboarding interview agents
• voice agents
• QA/compliance agents
• internal operations agents

These agents need to:

understand → retrieve → reason → act → update systems → measure outcomes.

Not just generate text.

4. Voice AI

We expect voice agents to become an important interface across the company.

Potential use cases:

• inbound sales
• outbound lead reactivation
• learner onboarding
• learner check-ins
• customer research
• escalation interviews

You should understand or be capable of quickly learning:

• STT/TTS
• real-time streaming
• telephony
• latency
• tool calling
• conversation state
• CRM integration
• call evaluation

5. Evaluate the AI stack

New models, frameworks and AI products appear constantly.

We don't want to chase all of them.

You should be able to answer:

• What should we build internally
• What should we buy
• Which model should we use
• Does LangGraph/custom orchestration/etc. actually help here
• Does this need an agent at all
• What is the expected ROI
• What is the cheapest reliable architecture
• What breaks when this goes from 20 users to 2,000

You will act as the technical filter between hype and business value.

6. Lead a small high-output team

Over time you'll manage engineers/automation builders working underneath you.

But this is not a delegate everything position.

We expect the technical leader to remain close enough to the code and architecture to:

• review implementations
• unblock engineers
• prototype difficult pieces
• establish standards
• debug failures
• make architecture decisions
• train the team
• increase the team's engineering velocity

The Person We're Looking For

You are probably someone who has been a:

• Founding AI Engineer
• Founding Engineer
• Applied AI Engineer
• Senior/Staff AI Engineer
• AI Engineering Lead
• AI Systems Architect
• Principal Engineer at a smaller company

Titles don't matter much to us.

Evidence does.

You should be able to show us things you have actually built.

Strong signals

• Built production AI systems end-to-end
• Strong Python/backend fundamentals
• Agent/tool-calling systems
• LangGraph or equivalent orchestration
• RAG/retrieval systems
• Vector databases
• Evaluation frameworks
• APIs/webhooks/event-driven systems
• PostgreSQL
• Cloud deployment
• observability
• CRM/SaaS integrations
• voice AI experience
• Claude Code/Cursor/Codex-style AI-assisted development
• startup or zero-to-one experience

Even stronger signals

You can point to something and say:

I designed this, built the first version, shipped it, watched it fail, fixed it, and eventually got other engineers building on top of it.

That is much more interesting to us than certificates or knowing every AI buzzword.

What We Don't Want

We're probably not a fit if:

• you only manage engineers
• you primarily make architecture diagrams
• you've never shipped AI into production
• most of your experience is notebooks/models/research
• your agents are essentially prompt chains
• you need perfectly written product requirements before starting
• you optimise for sophisticated architecture instead of working systems
• you're uncomfortable operating in ambiguity

How You'll Be Measured

Not by the number of AI projects launched.

By outcomes:

• hours of human work removed
• operational errors reduced
• response time improved
• decisions improved
• adoption by the team
• reliability
• cost per workflow
• measurable commercial impact

The goal is simple:

Build AI systems that make the organisation materially more capable.

You will work directly with the founder and have substantial ownership over how the company's AI infrastructure evolves.

Comp : 25-30LPA