Themathcompany
AI Architect
Designs and delivers production-grade GenAI architectures — multi-agent systems, RAG pipelines, LLMOps/AgentOps — leading cross-functional teams from design through operations. Requires deep cloud-native expertise (AWS/Azure/GCP, Kubernetes) plus hands-on LLM tooling like LangChain, vector databases, and MLflow.
- Compensation
- Not disclosed
- Location
- Bengaluru
- Work setup
- Onsite
- Seniority
- Not specified
- Employment
- Contract
- Experience
- 3+ years
- Category
- Architecture
- Posted
Skills
Must have
- 10+ years software engineering experience with 3+ years in architecture or senior technical leadership roles
- Proven track record designing and delivering cloud-native, production systems at enterprise scale
- Hands-on experience with GenAI/LLM systems, RAG, NL-SQL, agentic frameworks or similar productionized AI applications
- Strong knowledge of system design patterns (microservices, event-driven, CQRS, hexagonal architecture) and low-level design patterns
- Experience integrating ML/LLM services with enterprise data platforms and APIs while meeting security/compliance requirements
- Solid engineering background in at least two of: Python, TypeScript, Go, Java, C#
- Cloud & infra: AWS/Azure/GCP, Kubernetes, Docker, serverless (Lambda, Functions, Cloud Run), GPU instances
- GenAI & ML: Hugging Face Transformers, OpenAI APIs, LangChain, LlamaIndex, Semantic Kernel, Haystack
- Vector stores: FAISS, Pinecone, Weaviate, Chroma, Postgres+pgVector
- LLMOps/MLOps: custom ops pipelines, MLflow, TFX, BentoML, Kubeflow
- Data & integration: Kafka, Spark, Airflow, Flink, ETL/ELT, data lakes, API gateways (Apigee)
- DevOps & IaC: Terraform, Pulumi, CloudFormation, GitHub Actions, Jenkins
- Observability & security: Prometheus, Grafana, OpenTelemetry, Jaeger, ELK, Datadog; Vault, IAM, OAuth2/OIDC/SAML, Snyk, SonarQube, SAST/SCA, OWASP/CWE/CVE
- Databases & storage: relational (RDS/Cloud SQL), NoSQL (MongoDB, DynamoDB, Cosmos DB), Redis, S3/Blob/GCS
- Compliance knowledge: GDPR, HIPAA, SOC2; responsible AI practices
Nice to have
- Experience operating LLMs in regulated industries (pharma)
- Familiarity with prompt auditing, hallucination detection, and automated quality checks
- Background in knowledge engineering, semantic search, or knowledge graphs
- Academic background in CS, ML, or equivalent applied experience
- Mobile & cross-platform basics: React Native, Flutter
Full description
Job Description
Were looking for a seasoned Software Architect with deep expertise in cloud-native enterprise systems and Generative AI. You will define and deliver scalable, secure, and production-grade GenAI architectures — including multi-agent, RAG, LLMOps and AgentOps systems — and lead cross-functional teams to build and operate them. This role combines hands-on technical leadership, systems thinking, and strong stakeholder management.
Key responsibilities
Architecture & System Design
- Design scalable, modular, and cloud-native architectures for GenAI applications(microservices, event-driven, serverless).
- Define system boundaries, data flows, orchestration, and integration patterns forLLMs, vector DBs, embedding services, and tool integrations.
- Produce architecture artifacts ( Layered Architecture Diagrams, C4 Models, DFDs,Class, Sequence, ER & Use Case diagrams, different types of blueprints, APIcontracts, design and trade-off decisions).
- GenAI & Agentic Systems
- Architect and deliver Retrieval-Augmented Generation (RAG) pipelines, NaturalLanguage to SQL Flows, fine-tuning strategies, multi-modal capabilities, and tool-augmented agents.
- Design agent orchestration and multi-agent frameworks enabling planning,reasoning, and secure tool invocations, implement and design Agent prototypes andCommunication Protocols.
- Define prompt engineering standards, memory models(episodic/semantic/procedural), and context management.
LLMOps & AgentOps
- Define and implement model lifecycle pipelines: training, fine-tuning, validation,deployment, rollback, and monitoring.
- Build AgentOps processes for agent lifecycle, behavior tracking, governance andperformance optimization.
- Automate CI/CD for models, agents and services (MLflow, TFX, BentoML, custompipelines).
Integration, Security & Compliance
- Integrate GenAI services with enterprise systems (ERP, CRM, data lakes, APIs) usingsecure, scalable interfaces.
- Ensure secure access controls, data privacy, encryption, and compliance (GDPR,HIPAA, SOC2).
- Define responsible AI practices: bias mitigation, explainability, audit trails, andoutput governance.
- GenAI security — classify, encrypt & sign data/models; enforce least-privilege withshort-lived creds and CI/CD security gates; telemetry, drift/hallucination alerts, kill-switch & runbooks.
- Agentic AI security — provable agent identity/attestation, tool allowlist + human gatefor high-risk actions; ephemeral scoped tokens, sandboxed execution,and replayable audit traces.
Observability, Ops & Cost Optimization
- Define telemetry, tracing, and logging for models and agents; monitor performance,drift, hallucination rates and user feedback loops.