[Hiring] AI/ML Architect @66degrees

Role Description

An AI/ML Architect at 66degrees designs and ships production AI systems for enterprise clients and leads the engineers who build them. The Architect is the senior technical person on an engagement: the primary technical point of contact for the client, the owner of the architecture, and the person the team takes direction from.

The role is hands-on and client-facing in equal measure. Architects take a business problem, define the solution and the value it delivers, and then build and deploy it. They write and review code, make the architectural decisions, and stay with the system through production. They also run discovery, present to client executives, and are the face of 66degrees on the engagement.

This role is based in India and works daily with US-based colleagues and clients. Engagements typically run 12 to 16 weeks and are delivered by a Delivery Manager, one or more Architects, and a team of engineers.

Responsibilities

• Build
  • Design the architecture for each engagement: data pipelines, classical ML, agentic and generative AI, and the production path on Google Cloud, with the trade-offs written down.
  • Write and review the code that matters: core logic, orchestration, evaluation, the parts an engineer shouldn't be guessing at.
  • Take systems to production and own them through handoff: deployment, monitoring, drift, retraining, incident response, and a client who can run it after we leave.
  • Choose the right approach for the problem, and defend it on cost, explainability, and fit rather than on what's fashionable.
• Lead the team
  • Direct a team of engineers: write the brief with the reasoning, not only the tasks; confirm it landed before code is written; review the work.
  • Run a team where engineers raise problems early and are heard when they do.
  • Develop engineers: pairing, giving them the client call, making them present the design review, and telling the Delivery Manager who is ready for more.
  • Delegate real work and still own the hard parts. Not a solo builder; not a delegator who only checks in.
• Own the client relationship
  • Run discovery: turn an ambiguous ask into a scoped plan with success criteria the client agrees to.
  • Hold the solution to what the client actually needs, and when scope and need diverge, raise it early with options and what each costs.
  • Raise risk the day you see it, up and across, in business terms, with options. Report "on track" only when it is.
  • Present to client executives: the design, the trade-offs, the status, and the recommendation, answer first.
  • Spot expansion: where the client's next problem is, and tell sales.

Qualifications

• Technical (required)
  • Has personally designed and shipped production ML or AI systems, more than once, and owned them after launch.
  • A data science foundation with range: several technique families (classification, forecasting, NLP, deep learning, recommendation, anomaly detection) each tied to a real business outcome.
  • Has built a generative or agentic AI system that real users depended on.
  • Production fluency on a hyperscaler: Google Cloud preferred (Vertex AI, BigQuery, Cloud Run or GKE); deep AWS or Azure experience with a credible translation is acceptable.
  • Strong software engineering: expert Python and SQL, containers, CI/CD, and the habits that make code someone else can run.
  • Currently hands-on, with recent, specific examples of code written and architectural decisions made.
• Leadership and delivery (required)
  • Currently leads a team of engineers or data scientists, delegating real work while owning the architecture and core logic.
  • A repeatable method for taking an engagement from discovery through handoff: phases, milestones, who does what, validation before anything ships.
  • A track record of reshaping scope when the original ask wasn't the right solution, and of making the business case for the change.
• Communication and judgment (required)
  • Answers the question asked, first, then the detail. Adjusts depth for the audience.
  • Executive presence: speaks with conviction, holds a position under challenge.
  • Raises risk before it becomes a problem, to the person who can act on it, with a proposed fix.
  • Comfortable operating with autonomy across a significant time-zone gap with US colleagues and clients.

Requirements

• Bachelor's degree in computer science, data science, engineering or a related field, or equivalent experience.

Preferred

• Consulting or professional services background, with fluency in SoWs, scope, and billable delivery.
• Google Cloud certifications, or delivery alongside Google teams.
• Experience with agentic frameworks and patterns in production.
• Regulated-industry delivery (financial services, healthcare) where explainability and audit are requirements.
• Daily use of AI coding assistants and agents in your own work.
• Experience working with a US-based team from India, and a point of view on how to make it work.

What the job demands

• The client sees you. You present the design, you deliver the bad news, you defend the recommendation.
• The plan will be wrong sometimes. Client problems rarely arrive fully defined, and scope moves.
• Your engineers learn from you. If you raise risks, they will.
• The US is asleep for half your day. You will make decisions with incomplete direction and own them.
• It's hands-on. If your week has become meetings and status calls, this is not the role.