Job Description

We are looking for a hands-on Tech Anchor / Senior Software Engineer to drive the design and implementation of core AI capabilities within our Intelligent Data Analytics Platform (Lightspeed), spanning multi-agent orchestration, natural language-to-SQL generation, and semantic data discovery. This role is for a strong individual contributor who can operate at both architectural and implementation depth — someone who anchors the team technically by writing production-grade code, solving the hardest problems, and setting engineering standards through example.

You will be a key contributor to an AI-first platform that enables users to explore, query, and analyze enterprise BigQuery data through natural language — reliably, accurately, and at scale.

1. Architecture & System Design
- Contribute to the design of scalable, multi-agent AI architectures for data discovery and query generation
- Design components and modules across agent orchestration, tool systems, and LLM integration
- Evaluate trade-offs across design choices (e.g., single vs multi-agent, RAG vs fine-tuning, deterministic vs probabilistic pipelines)
- Participate in design reviews and contribute to architecture decision records (ADRs)

2. Hands-On Engineering & Execution
- Write production-grade code across agent frameworks, backend APIs, and frontend interfaces daily
- Build and evolve reusable AI components (agent tools, embedding pipelines, evaluation frameworks)
- Implement LLM-powered workflows including NL-to-SQL generation, semantic search, and metadata enrichment
- Develop services enabling intelligent data access (vector search, hybrid retrieval, query scope management)
- Implement guardrails, validation layers, and observability for AI-generated outputs

3. Full Stack Development
- Build performant backend services (Python/FastAPI) and interactive frontends (Angular/React) for data exploration
- Develop both conversational (chat) and structured (API) interfaces for analytics
- Build evaluation and benchmarking tooling for continuous AI quality measurement
- Own features end-to-end from design through deployment and monitoring

4. Semantic Search & Embeddings
- Implement vector embedding pipelines for metadata discovery (pgvector)
- Build semantic retrieval across datasets, tables, and columns with hybrid search strategies
- Optimize search relevance through embedding strategies, re-ranking, and evaluation metrics
- Contribute to data quality and governance capabilities within the platform

5. Engineering Excellence
- Write clean, maintainable, and scalable code following best practices (SOLID, DRY, design patterns)
- Actively participate in code reviews and set quality standards through your own contributions
- Perform root cause analysis on agent failures and implement systematic fixes
- Anchor the team technically — be the go-to person for complex implementation challenges

6. Collaboration
- Partner with Product, Data Engineering, and Platform teams on feature delivery
- Support teammates through pair programming, knowledge sharing, and technical guidance
- Contribute to sprint planning, estimation, and technical feasibility assessments
- Help onboard new team members and share domain expertise

What We're Looking For

- 5+ years of professional software engineering with strong hands-on coding ability
- Experience building AI-powered applications or working with LLM-based systems in production
- Ability to take ambiguous requirements and deliver working, tested software independently
- Strong debugging and problem-solving skills across the full stack
- Track record of owning and delivering complex features end-to-end

---

Technology Stack

- Programming: Python (primary), Java, TypeScript, Angular/React
- AI/ML: Google ADK, LangChain/LangGraph, OpenAI/Gemini APIs, prompt engineering, RAG pipelines
- Data & Cloud: GCP (BigQuery, Vertex AI, Cloud Run preferred)
- Backend: FastAPI, Pydantic, SQLModel/SQLAlchemy, PostgreSQL (pgvector)
- Frontend: Angular or React, TypeScript
- CI/CD & Infra: Terraform, GitHub Actions, Docker
- Evaluation: Custom eval frameworks, LLM-as-judge patterns
- Tools: Git, Alembic

---

Nice to Have

- Experience with Google Agent Development Kit (ADK) or similar agent frameworks (AutoGen, CrewAI, LangGraph)
- Exposure to NL-to-SQL or text-to-code generation systems
- Knowledge of ML fundamentals — embeddings, classification, clustering, evaluation metrics
- Experience with vector databases and semantic retrieval optimization
- Familiarity with data governance (metadata management, lineage, data quality)
- Experience building developer tooling or platform SDKs