GenAI engineers, data scientists and architects screened for a production track record — RAG, agents, MCP — including engineers who have shipped LLM systems for banks, where AI has to survive compliance. First CVs in 48 hours.
Everyone "works with AI" now. The ladder keeps the meaning in that phrase — we only present profiles from AI Developer upward: people who build systems, not people who use them.
CVs hide the gap between "did a GenAI demo" and "runs GenAI in production." We screen for continuity and depth — nothing else counts.
Banks want assistants and agents. Shipping them inside a bank is a different discipline — these four constraints are our screening bar:
Agents that aggregate across core systems without crossing a single permission boundary.
Answers built only from data the user is allowed to see — enforced in architecture, not in the prompt.
An assistant that answers questions must refuse to move money. Guardrails are systems design.
In banking a wrong answer is a regulatory event — logging, evaluation and traceability ship with v1.
Hires Data Scientists and AI Architects through NDS — body leasing and permanent, with more AI profiles in the pipeline.
A long-running delivery relationship, currently calling for AI and AML profiles. When a firm that audits everyone buys AI engineers, screening rigor is the product.
Most vendors describe their bench; here is a page of it. Published with the candidate's consent — identity disclosed at introduction, everything else real:
| Period | Role | What was built |
|---|---|---|
| 2023 – now | Data/AI Architect & Tech Lead · NDS | LLM & RAG solutions for banking, finance and consulting (incl. HSBC, Swedbank, PwC delivery track record); AI agents (CrewAI, LangGraph); hybrid search for financial data; document understanding; model fine-tuning. Azure AI Search, GCP Vertex, AWS. |
| 2021 – 23 | Data Science Tech Lead · TV broadcaster | Multimodal Smart Search AI across video, music, podcasts, audiobooks — GCP, PyTorch, Elasticsearch, MLFlow. |
| 2020 – 21 | Lead Data Scientist · AI fintech | Deep-learning algotrading platform — time-series Transformers, HPC on SLURM, ensemble optimisation. |
| 2008 – 09 | Algorithm Researcher · Netflix Prize | 2nd place in the $1M contest — The Ensemble team; large-scale recommender systems. |
| 2009 – 10 | Senior Backend / Data Miner · Polar Rose | Face detection & recognition at scale — company acquired by Apple. |
From SAS consulting through NLP research to production GenAI — the full arc, not a bootcamp certificate.
University of Warsaw, MIM faculty — thesis on automatic text segmentation, long before it was called AI.
Banking and consulting engagements where a model's error is a compliance event — the context our screening tests for.
Full blind CV available on request — hello@newdigitalstreet.com. Employers named appear as published in the candidate's consented profile.
The same commitments as all our staffing. Most vendors won't put theirs in writing.
Tell us what you're building and at which ladder step you need people. First anonymized profiles within 48 hours — and if the role as briefed won't attract the level you want, we'll say so at the start.
Get AI Engineer CVs in 48hContinuity and depth. We look for an engineering foundation that predates the AI wave — minimum four years in IT, ideally Python-heavy — plus at least a year of hands-on work with agent frameworks, RAG systems and MCP that the candidate can walk through decision by decision. Demos are ignored; architecture conversations are not.
Python-centric: LangChain, LangGraph and CrewAI for agents; RAG with hybrid search; Model Context Protocol; Pandas, NumPy, SciPy and Spark on the data side; Azure, GCP and AWS AI services in production. Other stacks on request — the screening method is the constant.
First matching CVs within 48 hours — the same contractual SLA as all our staffing. AI roles at Expert and Solution Architect level can take longer to close; we tell you the realistic timeline at the brief, not after three weeks.
Yes — including engineers who have built LLM and RAG systems for banks and financial institutions, where data governance and guardrails are part of the definition of done. That experience transfers well to any regulated environment.
Both. Most AI engagements start as staff augmentation for speed, with try-&-hire conversion when the fit is proven. Permanent recruitment runs through our IT Recruitment service with the same screening ladder.
Put a slot in a calendar, or send the brief and we will come back to you. Either way you are talking to a partner, not a queue.
Thirty minutes, no deck. Whoever's calendar suits you — they take the call themselves.
Roles, platform, headcount, timeline. A link to a brief is even better.