Senior Agentic AI Engineer

About Infocion

• Infocion is a US-based Data & AI consulting firm delivering transformation across Life Sciences, Pharma, Biotech, and broader Healthcare Systems (HCLS). We help global clients unlock business value with next-gen AI—from advanced analytics to Agentic AI systems that drive automation, compliance, and growth.

Role Summary

• We're building a world-class Center of Excellence (CoE) in India and hiring multiple hands-on Agentic AI Engineers to design, build, and operate production-grade multi-agent applications.
• You'll convert clinical and operational use cases into reliable agentic workflows, integrate with real systems (EHR/FHIR, CTMS, CRM), and ship secure solutions under healthcare-grade governance.
• You will also help build our AgenticOps platform that provides orchestration, governance, and observability for safe multi-agent operations.

What You'll Do

1) Build Agentic Systems

• Design and implement multi-agent workflows (reason/plan/act) with tools, memory, and retrieval (RAG/graph) for HCLS use cases (e.g., prior auth, trial matching, patient timelines, protocol/safety narrative automation).
• Integrate with EHR/FHIR/HL7, CTMS, CRM, and other operational systems; develop robust API adapters and event pipelines.
• Engineer prompt/tool/policy layers (function calling, MCP, guardrails) for reliability, cost, latency, and safety.

2) Contribute to our AgenticOps platform

• Build platform capabilities: agent scheduling & orchestration, sandboxing/isolation, policy-as-code guardrails, red/blue-green/canary rollout, versioning, and rollback.
• Implement observability & governance: per-agent tracing, lineage/provenance, evaluation harnesses, drift/bias checks, audit dashboards, and PHI-safe logging.
• Create internal assets: reference architectures, maturity checklists, AgenticOps runbooks, and ROI calculators.

3) Models, Data, and MLOps

• Evaluate and integrate LLMs (GPT-4o, Claude 3, Gemini 2.5, Llama 3.x, OSS variants), including parameter-efficient tuning where needed.
• Implement RAG/knowledge orchestration with VectorDB/GraphDB (Pinecone, Weaviate, Milvus, FAISS, pgvector, Neo4j).
• Own CI/CD for agents and models (Docker/Kubernetes), testing, telemetry, and SLOs.

4) Delivery & Collaboration

• Turn ambiguous requirements into MVPs and production rollouts; run discovery/demos with US stakeholders.
• Mentor juniors within the India CoE; enforce engineering best practices (tests, benchmarks, postmortems).

Must-Have Qualifications (Hands-on Engineer Level)

• 5–7 years total software/ML engineering experience, including 2+ years building LLM/agentic applications in production.
• AgenticOps platform experience (required): 1+ year building or operating an agent orchestration/governance platform (or equivalent hands-on experience composing multi-agent production systems) covering: scheduling, isolation, policy-as-code, evaluation gates, progressive delivery (canary/blue-green), observability, and audit/rollback.
• Hands-on with agent frameworks:
  Agno 2.0, LangChain (or AWS Bedrock AgentCore, Vertex AI Agents, Azure AI Foundry).
• Strong LLM
  integration & orchestration: prompt/program design, tool/function calling, context/memory, safety filters, MCP.
• RAG & knowledge layers with VectorDB/GraphDB; data ingestion, chunking, metadata, hybrid search, evals.
• MLOps & Ops:
  Docker/K8s, CI/CD, tracing/telemetry (Langfuse/OpenTelemetry/W&B), cost & latency tuning, rollback strategies.
• APIs & data plumbing: REST/GraphQL, Kafka/event streams; secure integration patterns.
• Awareness of HCLS compliance (HIPAA, 21 CFR Part 11, GxP) and PHI-safe retrieval design; you know what must be logged and why.
• Clear communication; ability to demo and explain trade-offs to technical and non-technical stakeholders.

Core Tools & Technologies (you'll touch many of these)

• Agentic frameworks:
  Agno 2.0, LangChain; plus AWS Bedrock AgentCore, Vertex AI Agents, Azure AI Foundry
• AgenticOps (
  platform) concepts: orchestration/scheduling, circuit breakers, feature flags, canary/blue-green, policy-as-code, sandboxing & isolation, meta-evaluation gates
• LLMs:
  GPT-4o, Claude 3, Gemini 2.5, Llama 3.x; open-source selection & tuning
• Knowledge layer: Pinecone, Weaviate, Milvus, FAISS, Postgres + pgvector, Neo4j/Graph
• Data & integration:
  REST/GraphQL, Kafka; FHIR/HL7 connectors; EHR/CTMS/CRM adapters
• Cloud & deployment: Docker, Kubernetes; AWS (Bedrock/SageMaker), Azure (OpenAI/Fabric), GCP (Vertex)
• Observability & ops: Langfuse, OpenTelemetry, Weights & Biases; agent tracing, audit dashboards
• Security & compliance: HIPAA, SOC 2, PHI handling, audit loggin

Compensation

• Pre-revenue stage:
  $200/month retainer (payment deferred until first revenue), flexible 4–40 hrs/week.
• Post-revenue stage
  :  $25 to $35 per hour (depending on experience level), continues until successful completion of 2 or 3 major projects.
• Full-time stage:
  Performance–based Eligibility after 2 or 3 large projects
• Employment with fixed salary of ₹30 to ₹45 LPA  (depending on experience and prior performance)
• Delivery Bonus: up to ₹1.5 L per project (with a yearly/cap)
• Annual Bonus: up to ₹2.5 L (performance-based, tied to client satisfaction).
• Benefits: Health insurance, 15 days paid leave, PF contributions