Python AI/ML Developer

We’re looking for a hands-on Python engineer with deep experience in building GenAI applications at production scale. You’ll design and implement LLM-powered systems—prompt pipelines, RAG, agents, evaluation frameworks—and ship reliable, secure, and cost-efficient AI features used by customers/stakeholders.

Key Responsibilities

• Design & Build GenAI Services: Implement prompt orchestration, tool-using agents, and RAG pipelines (indexing, chunking, embeddings, retrieval, ranking, grounding, guardrails).
• Model Integration: Integrate OpenAI/Azure OpenAI/Claude/Llama (hosted or self-hosted) via SDKs; evaluate models for latency, accuracy, and cost; fine-tune or instruct-tune where appropriate.
• Data & Features: Build ETL/data pipelines to prepare un/structured data; manage vector databases; implement prompt templates and prompt versioning.
• Evaluation & Observability: Define offline metrics (precision@k, hallucination rate, factuality) and online metrics (success rate, latency, CX/CSAT); set up tracing and telemetry (e.g., LangSmith, Weights & Biases, OpenTelemetry).
• Productionization: Write robust, tested APIs/microservices; containerize; CI/CD; ensure security, compliance, and privacy (PII handling, content filters, RBAC).
• Collaboration: Work closely with Product, Data, Security, and MLOps to move from POC → MVP → production.
• Cost/Latency Optimization: Token optimization, model selection, caching, batching, prompt compression, spec exec.

Must-Have Qualifications

• 5–10 years of professional software development with Python (fastAPI/Flask, asyncio, typing, packaging).
• 2+ years hands-on with LLMs/GenAI: prompt engineering, RAG, embeddings, function calling/tools, guardrails.
• Experience with one or more LLM frameworks: LangChain, LlamaIndex, Haystack, or custom orchestration.
• Vector DBs: Pinecone, FAISS, Chroma, Weaviate, pgvector, or Milvus.
• Model Providers: OpenAI/Azure OpenAI, Anthropic Claude, Google Gemini, Meta Llama, Mistral, or Cohere.
• Cloud & DevOps: Deploy on Azure/AWS/GCP, Docker, CI/CD (GitHub Actions/GitLab), logging/monitoring.
• Strong grounding in data structures/algorithms, HTTP APIs, authN/Z, and unit/integration testing.
• Practical knowledge of prompt testing & evaluation (A/B tests, golden sets, LLM-as-judge with safeguards).

Good-to-Have / Preferred

• Fine-tuning or LoRA/QLoRA, PEFT, or domain adaptation for LLMs.
• Multimodal (image/doc QA, speech) or agents (planning, tool-use, code exec).
• Information retrieval (BM25, hybrid retrieval, re-ranking with bge, E5, ColBERT).
• MLOps: Feature stores, model registries, Ray/Databricks; Airflow/Prefect for orchestration.
• Security/Compliance: SOC2, ISO 27001, GDPR, content moderation, PII redaction.
• Domain experience: Enterprise search, Knowledge assistants, Customer support automation, Document intelligence, Coding copilots, Marketing content generation.