Job Title: AI Engineer
Location: Remote (offshore – India) working IST hours with 3 or 4 hours overlap with EST
Duration: 6+ Months, long term
Start: 05/18
They have internal SMEs and Data Managers who will partner on this. Hands on technical is more important here to get in the weeds and do the dev work. Would be more of a nice to have for the K to be able to gather requirements / business facing. OCR exp and RAG pipeline would be very strong plusses – would accept framework and Python exp. Databricks, Snowflake and AWS are their platforms. Jupyter notebooks is a must.
Key Responsibilities
- Conversational Agents: Design and implement LLM-powered agents with tool/function calling, dialogue state management, and robust guardrails and safety controls; independently drive solutions from prototype to deployment.
- RAG Pipelines: Build end-to-end retrieval pipelines (indexing, chunking, embeddings, reranking, hybrid retrieval) to deliver high-precision, grounded answers; proactively diagnose issues and implement fixes.
- Document AI: Ingest and parse complex PDFs using OCR and multimodal LLMs for image understanding, including table extraction and layout-aware parsing to capture structure and context; resolve parsing challenges with creative approaches.
- Prompt Engineering: Develop and iterate on prompts, templates, and few-shot exemplars; run A/B tests and automated evaluations for quality and robustness; propose improvements without needing direction.
- Evaluation & Operations: Measure grounding, hallucination, latency, and cost; deploy, observe, and monitor services in cloud environments with clear SLIs/SLOs; take ownership of incident resolution and preventive actions.
- Project Documentation & Communication: Create and maintain requirements, technical designs, test plans, runbooks, release notes, user guides, risk/issue logs, change records, status updates, meeting notes, incident/outage notices, and decision summaries with minimal oversight.
Skills Essential
- Business Analysis & Stakeholder Engagement: 3+ years leading client-facing workshops, requirements sessions, and solution discussions; able to independently clarify needs and align decisions.
- Applied ML/AI: 3+ years building ML/AI applications, including 1–2+ years hands-on with LLMs and RAG; demonstrated ability to own problems end-to-end and deliver working solutions.
- Programming: Strong Python; experience with LangChain/LlamaIndex and Hugging Face; writes production-quality code with limited guidance.
- Retrieval & Ranking: Hands-on with vector databases (e.g., Qdrant, Milvus, Elasticsearch), semantic search, and rerankers; proactively evaluate trade-offs and tune for relevance.
- Chunking & Embeddings: Practical expertise in chunking strategies (semantic, layout-aware) and embedding selection/tuning; proposes data and model changes to improve quality.
- Document Processing: Experience with table extraction, layout parsing, OCR, and multimodal LLM-based image understanding and structure detection; resolves edge cases independently.
- Tool Integration: Experience integrating tools and data sources via MCP or similar agent-tool frameworks; comfortable designing and wiring new tools without supervision.
- MLOps & Cloud: Solid prompt engineering and cloud deployment (AWS/Azure/GCP), containers, and CI/CD; monitors, troubleshoots, and optimizes services autonomously.
Desirable
- Awareness of Emerging AI: Sound understanding of developments in generative and agentic AI to help frame innovative aspirations and practical roadmaps; proactively identify opportunities and risks.
- Experience in the pharmaceutical industry.
