AI/ML Solution Architect

Posted 2hrs ago

Employment Information

Education
Salary
Experience
Job Type

Report this job

Job expired or something wrong with this job?

Job Description

AI/ML Solution Architect designing secure, scalable AI assessment platforms for Pratham International’s global education programs. Architecting RAG, LLM judging, OCR, cloud, and LLMOps systems.

Responsibilities:

  • Own solution design from initial workshop through delivery-ready blueprint
  • Lead solution workshops with programme, ministry, engineering, and external stakeholders
  • Convert requirements into service boundaries, data stores, sequence diagrams, decision gates, and open-question lists
  • Write architecture decision records and resolve conflicting technical decisions
  • Architect LLM generation, curriculum knowledge-graph retrieval, LLM-as-judge pipelines, and constraint-based paper assembly
  • Design human-AI workflows with auditability, segregation of duties, cycle-pinned configuration, and protection of student PII
  • Choose and govern models, prompts, evaluation gates, fallbacks, and OCR providers
  • Shape Azure/AWS platform architecture including identity, networking, LLMOps/MLOps, observability, and sovereign hosting
  • Define contracts, schemas, metadata logging, evaluation harnesses, release gates, rollback, cost controls, and FinOps guardrails
  • Stay with systems through build and first production cycle
  • Produce POVs, architecture packs, sequence packs, workshop artefacts, and reusable country adaptation patterns
  • Mentor technical leads on boundaries between model-based logic and deterministic workflow logic

Requirements:

  • 8+ years building software or data/AI systems, with recent experience as a Solution Architect, AI/ML Architect, or Principal/Staff Engineer with end-to-end architecture ownership
  • Shipped at least one production AI/ML system at scale that passed real security, cost, and operations review
  • Ability to produce delivery-ready architecture blueprints covering services, data stores, IAM, failure modes, evaluation strategy, and cost model
  • Deep cloud experience on Azure and/or AWS
  • Experience defending landing-zone and private-endpoint architecture and real compute/inference/OCR spend
  • Fluent in Python
  • Hands-on experience with classical ML, model training, evaluation, feature pipelines, LLM APIs, embeddings, and vector search under production load
  • Comfortable running workshops with non-engineers and translating between technical and non-technical stakeholders
  • Strong written English
  • Flexibility to work across time zones, including regular early mornings or late evenings
  • Strongly preferred: Azure AI plus AWS or GCP; LLMOps/MLOps; document AI/OCR at volume; structured-knowledge RAG and hybrid search; education, assessment, public-sector, health, finance, or identity domains; relevant cloud certifications; agent security; FinOps
  • Nice to have: Microsoft Fabric, Databricks, Semantic Kernel/Microsoft Agent Framework, MCP, constraint solvers, exam-blueprint/item-bank systems, multilingual generation, golden/eval datasets, government data systems, data-sharing agreements, or sovereign hosting experience

Benefits:

  • Impact at scale supporting learning journeys of millions of children
  • Autonomy to experiment with the latest SOTA models
  • Support from a 5-year strategic roadmap and funding from partners like the Gates Foundation and Anthropic
  • High-calibre, mission-driven global team
  • Inclusive and collaborative environment
  • Equal opportunity employer
  • Remote-first work arrangement
  • Travel to client workshops as needed