Logo
Kces

Data Scientist (LLM) / AI Research Engineer

Kces, Convent Station, New Jersey, us, 07961

Save Job

Data Scientist (LLM) / AI Research Engineer Base pay range: $140,000.00/yr - $160,000.00/yr

Are you passionate about building production-grade LLM systems, agentic workflows, and retrieval pipelines that power real-world applications? We’re looking for a hands‑on AI Research Engineer / Data Scientist to lead the design, development, and deployment of cutting‑edge generative AI solutions. In this role, you’ll own the full lifecycle — from proof‑of‑concept to production — while driving architecture, evaluation, scalability, and safety.

What You’ll Do

Lead POC → pilot → production lifecycle for LLM and agentic solutions (tool calling, planning, fallbacks).

Architect advanced retrieval stacks (chunking, hybrid search, re‑ranking) and hallucination mitigation strategies.

Build evaluation frameworks (offline + online, golden sets, CI/CD gates) for correctness, safety, and bias.

Implement confidence scoring, calibration, abstention policies, and user‑visible citation mechanisms.

Optimize cost, latency, reliability (caching, batching, routing, distillation/quantization).

Deploy secure, observable APIs/services with RBAC, tracing, and audit logging.

Own model/vendor selection, prompting vs. fine‑tuning decisions, and roadmap contributions.

Mentor junior team members and align delivery with stakeholder KPIs.

Must‑Have Experience

4–8+ years in applied ML, data science, or engineering with shipped LLM products (RAG, tool calling, QA, structured extraction).

Proven track record designing and integrating evaluation strategies and CI/CD gates.

Hands‑on experience with confidence calibration and abstention/deferral techniques.

Expertise in retrieval systems (hybrid search, filters, re‑ranking, chunking) and hallucination mitigation.

Strong Python engineering (FastAPI), SQL, containerization, and observability (logs/spans/metrics).

Cloud and vector search experience (Azure/AWS/GCP, Elasticsearch, Pinecone, FAISS).

Nice to Have

Fine‑tuning/LoRA, multi‑agent orchestration, prompt routing.

Advanced uncertainty modeling, LLM‑as‑a‑judge techniques.

Model optimization (distillation, quantization) and GPU/CPU trade‑offs.

Safety and compliance expertise (red‑teaming, model cards, DPIA/PIA).

Domain knowledge in insurance, finance, or healthcare.

Seniority level Director

Employment type Full‑time

Job function Information Technology

Industries IT Services and IT Consulting

Location: Hoboken, NJ

#J-18808-Ljbffr