Technical Lead / AI Solutions Architect

• Ship headless automation end to end. PR-review agents, release-notes agents, analytics pipelines, from concept to production.
• Make the architectural calls. When an agent needs memory, when it needs tools, when it needs a context graph, and how to keep cost under control.
• Think in work removed. “What human effort does this eliminate,” not “which model do we plug in.”
• Runs technical client conversations in English independently
• Leads discovery and presales so the client knows what they’re paying for
• Mentors and onboards developers; runs technical interviews
• Holds several client tracks in parallel without losing context
• Writes clearly: case studies, blueprints, client emails, workshop scripts

Nice-to-Have:

• Deep understanding of how AI works. Not surface-level API knowledge. A real grasp of models, context, agents, and their limits, so you know why something breaks and where.
• Tests everything new proactively. Approaches, technologies, libraries. When something appears, you’ve already tried it and know if it’s worth a client’s time.
• Stays on top of the field. Tracking what ships and separating hype from substance is the job, not a bonus.

Responsibilities:

• Run discovery with new clients: understand the domain, map the problem, and scope a solution
• Design and ship agentic workflows and software automation that remove real client work
• Own client tracks end to end, from architecture through to production delivery
• Build and maintain AI infrastructure: gateways, tracing, telemetry, headless agents
• Support presales: shape proposals, join technical calls, and make the client confident in the plan
• Lead and mentor the developers on your tracks; run technical interviews and onboard new hires
• Evaluate new approaches, libraries, and tools, then bring the ones worth using into client work Write the artifacts that carry the work: blueprints, case studies, client emails, workshop material

About the project: A fullstack, technology-agnostic engineer at the intersection of architecture, presales, and client delivery. Not a ticket-closing role. You walk into an unfamiliar client domain, run discovery, find where it hurts, and close the gap with agentic and software solutions. You’re not tied to one stack or language; you pick the right tool for the problem and learn whatever the client’s environment demands.

• Decompose the problem. Split a client’s problem into steps and decide what an agent should automate versus plain software. Know where not to put an LLM when a few lines of code do the job better.