Job Information
Date Opened
12/09/2025
Job Type
Contract
Work Experience
5+ years
Industry
Government & Public Sector
Salary
400 - 500 outside
City
London
Province
Greater London
Country
United Kingdom
Postal Code
EC1A
The role
Were hiring a
Data Scientist with strong Generative-AI experience to design, build, and ship AI-powered tools end-to-end. Youll work in a small, multi-disciplinary team and take ownership from discovery to deployment: scoping use-cases, building prototypes, hardening them for production, and putting the right evaluation and governance around them.
What youll do
- Build GenAI tools end-to-end (independently): chat/assistants, document Q&A (RAG), summarisation, classification, extraction, and workflow/agent automations.
- Own evaluation & safety : create offline/online eval sets, measure faithfulness/hallucination, bias, safety, latency and cost; add guardrails and red-teaming.
- Productionise : package as services/APIs or lightweight apps (e.g., Streamlit/Gradio/React), containerise, and integrate via CI/CD.
- Data pipelines : design chunking/embedding strategies, pick vector stores, manage prompt/versioning, and monitor drift & quality.
- Model strategy : select and mix providers (hosted and open-source), fine-tune where its sensible, and optimise for cost/perf/privacy.
- Stakeholder enablement : translate problems into measurable KPIs, run discovery, document clearly, and hand over maintainable solutions.
- Good practice : apply data ethics, security and privacy by design; align to service standards and accessibility where relevant.
Tech youll likely use
- Python (pandas, PyTorch/Transformers), SQL
- LLM frameworks : LangChain, LlamaIndex (or similar)
- Vector DBs : FAISS / pgvector / Pinecone (or similar)
- Cloud & Dev : Azure/AWS/GCP, Docker, REST APIs, GitHub Actions/CI
- Data & MLOps : BigQuery/Snowflake, MLflow/DVC, dbt/Airflow (nice to have)
- Front ends (for internal tools) : Streamlit / Gradio / basic React
Must-have experience
- 7+ years in Data Science/ML, including hands-on delivery of GenAI products (not just PoCs).
- Proven ability to ship independently : from idea prototype secure, supportable production tool.
- Strong Python & SQL ; solid software engineering habits (testing, versioning, CI/CD).
- Practical LLM skills: prompt design, RAG , tool/function calling, evaluation & guardrails , and prompt/model observability.
- Sound grasp of statistics/experimentation (A/B tests, hypothesis testing) and communicating impact to non-technical audiences.
- Data governance, privacy and secure handling of sensitive data.
Nice to have
- Experience in regulated or public-sector-like environments.
- Azure OpenAI / Vertex AI / Bedrock; lightweight fine-tuning/LoRA .
- Front-end skills to craft usable internal UIs.
Please apply cv at
LNKD1_UKTJ
