New opportunity

Solution Architect

Anblicks · Ahmedabad, India

About this role

Anblicks seeks an AI Solution Architect to turn ambiguous business problems into practical AI solution blueprints, validate them through proofs of concept, and guide developers through implementation. The full-time position lists Ahmedabad and Dallas as locations.

Responsibilities

Lead business and technical discovery, assess whether AI is appropriate, define measurable outcomes, and prioritize opportunities by value, feasibility, risk, and adoption readiness.
Design platform-neutral options spanning generative AI, agentic workflows, retrieval-augmented generation (RAG), automation, predictive models, and hybrid approaches. Specify model interactions, orchestration, data and tool access, APIs, identity, observability, evaluation, security, and operational support.
Build or directly guide prototypes to test feasibility, user value, quality, performance, and risks. Support delivery through design walkthroughs, reference patterns, technical decisions, code-level guidance, and reviews.
Evaluate cloud AI services, foundation models, agent frameworks, integration patterns, and data platforms. Incorporate privacy, auditability, human oversight, content safety, and other responsible AI controls. Communicate decisions and trade-offs to business, engineering, risk, and executive stakeholders, and create reusable architectures, templates, scorecards, and playbooks.

Required qualifications

10+ years of technology delivery experience, including substantial solution architecture or technical leadership. Candidates need production AI delivery experience with LLM applications, RAG, agentic workflows, intelligent automation, or ML-enabled products; hands-on engineering in Python and/or a modern full-stack technology; and experience with APIs, cloud-native services, integration, testing, and deployment. The role also requires prompt and context design, model evaluation, grounding, tool integration, observability, secure deployment, enterprise data architecture, SQL, data quality, data-access controls, responsible AI, and clear stakeholder communication. Candidates must be able to define implementation increments, risks, and acceptance criteria, mentor engineers, and review architecture and code.

Preferred qualifications

Azure AI services, Azure AI Foundry, Azure OpenAI, Azure data services, or comparable platforms; Anthropic Claude, OpenAI-compatible APIs, Model Context Protocol, vector search, knowledge graphs, agent frameworks, and enterprise RAG; modern web frameworks, API gateways, containers, CI/CD, infrastructure as code, and production observability. Regulated-industry experience, familiarity with PII/NPPI controls, model validation, audit evidence, and human-in-the-loop approvals, or cloud or AI architecture certifications is also preferred.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Solution architectureGenerative AILarge language modelsRAGAI agentsIntelligent automationPredictive modelingPythonAPI integrationCloud-native servicesPrompt designContext designModel evaluationVector searchKnowledge graphsSQLData architectureData qualityData-access controlsObservabilitySecurityResponsible AIAzure AI FoundryAzure OpenAIAzure data servicesAnthropic ClaudeOpenAI-compatible APIsModel Context ProtocolAgent frameworksAPI gateways creating an incorrect grouped skill? no

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Turn ambiguity into measurable use cases
The role leads business and technical discovery, tests whether AI is appropriate, and prioritizes opportunities by value, feasibility, risk, and adoption readiness. Prepare to explain how you would define outcomes and acceptance criteria before choosing a solution.
Design end-to-end, platform-neutral architectures
The posting calls for options across generative AI, agents, RAG, automation, and predictive or hybrid approaches, with explicit design for data, tools, APIs, identity, evaluation, security, and operations. Prepare to explain the trade-offs behind an architecture choice.
Validate before scaling
Build or directly guide proofs of concept that test feasibility, user value, quality, performance, and key risks. Prepare an example of how prototype evidence would shape an implementation decision.
Carry architecture through delivery
This is a hands-on role: provide walkthroughs, reference patterns, technical decisions, code-level guidance, and reviews while mentoring engineers. The posting also calls for reusable templates, scorecards, and playbooks.
Build governance into the design
Embed privacy, security, auditability, human oversight, evaluation, and content-safety controls into architecture and delivery gates; communicate the resulting decisions to business, engineering, risk, and executive stakeholders.

Likely challenges to think through

Choosing when AI is warranted
A likely difficulty is balancing an attractive AI use case against simpler automation or non-AI options and uncertain adoption. Prepare a decision framework that compares measurable value, feasibility, risk, and readiness; this is a preparation hypothesis, not a stated employer problem.
Reconciling prototype speed with production evidence
A proof of concept may demonstrate promise without establishing quality, performance, security, or operational fitness. Prepare to describe evaluation thresholds, risks, and acceptance criteria that would justify—or block—scaled implementation.
Keeping flexibility without leaving implementation vague
Platform-neutral options must still become concrete choices for models, data access, integration, and support. Prepare to show how you would compare options against enterprise requirements, document trade-offs, and give developers actionable guidance.

Practice your answers

Technical practice

How would you decide whether an ambiguous business problem needs AI?

What to coverThe posting calls for discovery, measurable outcomes, and prioritization by value, feasibility, risk, and adoption readiness. Suggested preparation: explain what you would ask stakeholders, how you would compare AI with a simpler approach, and what evidence would justify proceeding.

How would you choose among RAG, an agentic workflow, automation, a predictive model, or a hybrid?

What to coverThe role involves platform-neutral options across these patterns. Suggested preparation: compare them against the problem, available data, required reliability, implementation complexity, and operational risk rather than assuming a generative AI solution.

What would an end-to-end blueprint for an enterprise AI solution contain?

What to coverThe posting specifies model interaction, orchestration, data and tool access, APIs, identity, observability, evaluation, security, and operational support. Suggested preparation: show how these parts fit together and identify decisions that must be validated before implementation.

How would you design and evaluate grounding for an enterprise RAG application?

What to coverRAG, grounding, enterprise data architecture, SQL, data quality, semantic concepts, and access controls are stated areas. Suggested preparation: discuss source selection, retrieval quality, permission-aware access, and how you would detect unsupported or outdated answers.

How would you control an agentic workflow that calls enterprise tools or APIs?

What to coverThe role covers agentic workflows, tool integration, identity, and human oversight; Model Context Protocol is preferred experience. Suggested preparation: address permitted actions, authorization, failure handling, traceability, and when an approval step is warranted.

How would you structure a proof of concept and decide whether it is ready to scale?

What to coverThe posting asks for prototypes that test feasibility, user value, quality, performance, and key risks. Suggested preparation: define measurable acceptance criteria, an evaluation approach, and the trade-off between a convincing demonstration and evidence of production readiness.

How would you compare foundation models and cloud AI services for an enterprise use case?

What to coverAssessing models, cloud services, agent frameworks, integrations, and data platforms is a stated responsibility; named vendors are preferred experience, not mandated choices. Suggested preparation: describe a platform-neutral scorecard covering quality, security, integration, performance, and operational fit.

How would you build responsible AI controls into the architecture and delivery gates?

What to coverThe posting names privacy, security, auditability, evaluation, content safety, human oversight, and model risk. Suggested preparation: explain where controls and evidence would sit in the design, including how sensitive-data access and potentially harmful outputs would be handled.

How would you operate and troubleshoot a deployed AI application?

What to coverObservability, secure deployment, testing, and operational support are stated requirements. Suggested preparation: discuss what you would monitor across model calls, retrieval, integrations, quality, and performance, and how those signals would guide incident diagnosis.

How would you turn an approved blueprint into implementation increments?

What to coverThe role requires implementation increments, risks, acceptance criteria, code-level guidance, and reviews. Suggested preparation: explain how you would sequence interfaces and tests, record technical decisions, and verify that delivery still matches the architecture.

Behavioural practice

Tell me about a time you brought clarity to an ambiguous business request.

What to coverThe role calls for leading discovery and defining measurable use cases. Suggested preparation: use a real situation, describe the questions and prioritization actions you took, and state the resulting decision or outcome.

Describe a time you explained a difficult architecture trade-off to different stakeholders.

What to coverThe posting names business, engineering, risk, and executive audiences. Suggested preparation: give the situation, how you adapted your explanation and handled disagreement, and the resulting decision.

Tell me about a time you guided engineers through a challenging implementation.

What to coverDeveloper enablement, mentoring, walkthroughs, and architecture and code reviews are stated responsibilities. Suggested preparation: describe the implementation context, the specific guidance or review actions you provided, and what changed as a result.

Describe a time a prototype or evaluation exposed a significant risk.

What to coverThe role uses proofs of concept to validate quality, performance, value, and risk before scaling. Suggested preparation: explain the situation, how you investigated and communicated the finding, and the resulting change in direction or scope.

Tell me about a reusable pattern you created from a delivery lesson.

What to coverThe posting calls for reference architectures, templates, guardrail patterns, scorecards, and playbooks. Suggested preparation: describe the original situation, what you made reusable and why, and the observed effect on later work.

Sharpen your knowledge

Topics to prepare

AI use-case selection
Required: discovery includes deciding whether AI is appropriate and prioritizing by value, feasibility, risk, and adoption readiness. For revision, take a hypothetical ambiguous business request, define a measurable outcome, compare an AI and non-AI option, and explain the priority you would assign.
Platform-neutral architecture patterns
Required: solution options span RAG, agentic workflows, automation, predictive models, and hybrids. For revision, compare two patterns for a hypothetical use case and sketch how model interaction, orchestration, APIs, identity, observability, and operational support would differ.
Enterprise data and grounding
Required: enterprise data architecture, SQL, data quality, semantic concepts, access controls, and grounding approaches. For revision, trace how a hypothetical answer would move from an authorized data source through retrieval or another grounding method, identifying quality and access-control failure points.
Evaluation and responsible AI gates
Required: model evaluation, privacy, security, auditability, human oversight, content safety, and risk controls. For revision, draft acceptance criteria for a hypothetical proof of concept covering answer quality, performance, sensitive-data handling, and a decision to proceed or stop.
Platform and model trade-offs
Required: assess cloud AI services, foundation models, agent frameworks, integrations, and data platforms against enterprise needs. For revision, make a short decision matrix covering capability, integration, security, and operating implications. Azure services and the named model/API options are preferred examples, not universal requirements.

Skills to revise

Scope a proof of concept
Required: build or directly guide prototypes that test feasibility, user value, quality, performance, and risks. As a preparation exercise, write a one-page plan for a hypothetical prototype with test users, measurable checks, implementation increments, and acceptance criteria.
Implement an API-backed AI flow
Required: hands-on Python and/or modern full-stack engineering, APIs, integration, testing, and deployment practices. As a preparation exercise, build a small mock-data flow with a model or tool API boundary, error handling, and tests; explain how you would deploy it securely.
Run a grounding evaluation
Required: prompt and context design, grounding, model evaluation, and observability. As a preparation exercise, assemble a small permitted test set, compare two prompt or context designs, record answer quality and failure cases, and specify what production signals you would monitor.
Review data access and SQL
Required: SQL, data quality, data-access controls, and secure deployment. As a preparation exercise, write a query over a hypothetical enterprise schema, identify missing or inconsistent records, and explain how the application would enforce access before exposing results to an AI workflow.
Guide an implementation review
Required: design walkthroughs, code-level guidance, architecture and code reviews, mentoring, and stakeholder communication. As a preparation exercise, review a hypothetical AI design for two technical risks and one delivery risk, propose changes, then summarize the trade-offs separately for engineers and executives.

Your preparation checklist

Draft a discovery and prioritization worksheet
For a hypothetical ambiguous business problem, write questions to test whether AI is appropriate, define measurable outcomes, and rank options by value, feasibility, risk, and adoption readiness. This is a practice exercise, not an Anblicks use case.
Prepare a production-delivery walkthrough
Select a real AI solution you worked on, if available. Summarize your architecture and engineering responsibilities, implementation, testing, deployment, measured outcomes, and lessons learned. Prepare a separate timeline of technology delivery and technical leadership experience relevant to the stated 10+ year requirement.
Sketch a platform-neutral solution blueprint
Compare two plausible approaches, such as RAG and a non-AI alternative. Diagram model interactions, orchestration, data and tool access, APIs, identity, observability, security, and operational support; explain when each approach would be appropriate.
Write a proof-of-concept evaluation plan
Specify a small prototype, acceptance criteria, and tests for feasibility, user value, quality, performance, and risk. Include how you would evaluate prompts, grounding, and tool use before recommending scaled implementation.
Map enterprise data and responsible-AI controls
Prepare a data-flow and control checklist covering SQL and data quality, access restrictions, privacy, auditability, human oversight, content safety, and secure deployment. Mark which decisions would need confirmation during discovery.
Show how you would enable developers
Prepare a sample implementation plan with increments, technical decisions, review checkpoints, and a code-level example you can explain. The role calls for hands-on guidance and architecture and code reviews, not just a design handoff.
Rehearse a platform trade-off briefing
Create a short decision scorecard comparing relevant models, cloud AI services, integration patterns, and data platforms against enterprise requirements. Explain the recommendation differently for engineering, risk, and executive audiences.
Bring questions about delivery expectations
Ask which AI opportunities are currently prioritized, how proofs of concept advance to production, what evaluation and governance gates apply, and how architects work with developers and stakeholders. These are questions to verify, not assumptions about Anblicks’ processes.

Availability last checked: 5 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in India. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…