Principal Software Engineer – Applied AI (Technical Lead)

About CENSUS

Key Responsibilities

• AI Application Architecture & Productization
• Own the reference architecture for LLM/AI features (service boundaries, APIs, data contracts, trust boundaries, deployment patterns).
• Convert prototypes into production-ready services with clear interfaces, type/schema discipline, and controlled dependencies.
• Define versioning strategy for prompts, tools, policies, and model endpoints (including rollbacks).
• LLM / Agentic Pipelines (Tool Use, RAG, Structured Output)
• Design and implement LLM pipelines for:
• Natural language → structured queries for database engines with guardrails.
• Retrieval-augmented generation (RAG) where appropriate (vector stores, hybrid retrieval, context management).
• Safe tool execution (bounded retries, timeouts, sandboxing, rate limits, strict output schemas).
• Implement schema-first output (e.g., Pydantic models / JSON schema) and robust parsing/validation.
• Reliability, Performance & Observability for Production Inference
• Own non-functional requirements: latency budgets, throughput, resilience, and cost control.
• Implement production patterns: caching, batching, streaming responses, backpressure/queues, circuit breakers, idempotency, and graceful degradation.
• Build observability: structured logs, traces, metrics, and error taxonomies that make failures actionable.
• Implement AI security controls (prompt injection defenses, tool access controls, data minimization, audit logging) aligned with client requirements.
• Collaboration & Delivery
• Translate requirements from client stakeholders and internal security / engineeringteams into robust AI-enabled solutions and data integrations
• Lead techncal delivery for AI components across client environments (cloud/on prem), aligning with security, compliance, and operational constraints.
• Own handover artifacts: architecture docs, runbooks, deployment guides, and support playbooks.
• Document solutions and participate in knowledge transfer with internal and partner teams.
• Technical Leadership & Mentorship
• Act as the technical lead for Applied AI delivery: raise engineering bar, mentor DS/engineers, and drive code quality.
• Own architectural reviews, refactoring plans, and prioritization of technical debt tied to reliability and delivery.

Minimum Qualifications

• BSc/MSc in Computer Science, Data Engineering, or a related field (or equivalent practical experience).
• 6+ years of software engineering experience, including technical leadership on production systems and AI-/ML-powered applications.
• Track record of making architecture decisions and driving cross-team execution.
• Strong Python engineering skills (services, tooling, testing) and experience building maintainable codebases.
• Demonstrated experience shipping LLM/AI-enabled applications into production.
• Proven ability to build robust APIs and services (FastAPI or equivalent).
• Experience with production reliability practices, including observability, incident response, performance tuning, and scalability.
• Comfortable integrating with big data and analytics platforms.
• Proven ability to define engineering standards and guide teams through architectural tradeoffs in ambiguous environments.
• Excellent written and verbal communication in English; able to lead technical discussions with engineers and client stakeholders.

Preferred /

• Experience with database query planning/optimization and guardrailed NL→SQL systems.
• Experience with vector databases / retrieval systems (FAISS or managed equivalents) and hybrid retrieval strategies.
• MLOps/serving experience: model gateways, inference optimization, caching strategies, cost controls.
• Experience with Docker/Kubernetes and deployments in restricted/on-prem environments.
• Security engineering exposure: threat modeling, secure coding, data handling requirements, and common AI security risks.

First 3–6 Month

• A clean, maintainable AI pipeline architecture with clear module boundaries and interface contracts (schemas, APIs, and data).
• Working evaluation harness + golden dataset with CI gating for regressions.
• Production-grade LLM features with measurable SLOs/SLAs: latency, correctness, reliability, and cost.
• Deployment-ready packaging for client environments (cloud/on-prem), with observability and operational runbooks.
• Team uplift: engineering standards, review discipline, and repeatable delivery patterns.

OUR Values & Core Competencies

Act with Integrity

Collaborate with Trust

Challenge with Curiosity

Innovate to Protect

Adapt with Precision