AI engineer

Established in 1997, Softeq was built from the ground up to specialize in new product development and R&D, tackling the most difficult problems in the tech sphere. Now we've expanded to offer early-stage innovation and ideation plus digital transformation business consulting. Our superpower is to deliver all of this under one roof on a global scale. So let's get started and build a better future together! We're looking for a Senior AI Engineer to join our team and contribute to its continued growth.

• Builds and hardens an AI assistant grounded on a curated data layer
• Delivers the product AI features: personalized recommendations, generated summaries and briefings, in-product assistance, learning suggestions, and detection of irregularities in reporting
• Owns prompt and context orchestration, retrieving the right data from the catalog while respecting the requesting user permissions
• Builds and maintains indexes for semantic search and keeps them current as the underlying tables change
• Implements guardrails: resistance to instruction injection through user input and through retrieved data, hard limits on accessible scope, output filtering
• Builds evaluation practice: example sets, regression runs when a model or prompt changes, and acceptance criteria
• Monitors inputs, outputs, quality, latency and cost per request, and supports explainability of answers
• Ships AI features into the service layer and the application interface

• Python plus real production experience integrating large language models
• Retrieval-augmented generation in practice: context assembly, relevance evaluation, working within context limits, vector indexes and keeping them fresh
• Databricks as a context source: reading from the catalog, honouring grants and row-level rules when placing data into a prompt, platform vector search, and invoking models from SQL and jobs
• Prompt and context engineering as a versioned, repeatable process
• Evaluation of AI features: test sets, metrics, regression when the model updates
• AI security: prompt injection, cross-tenant leakage, output filtering
• FastAPI for the service side, and enough React to land a feature in the interface
• Будет плюсом: Publishing and monitoring models as endpoints on the platform, with a single access point that handles limits and cost; Azure OpenAI or another managed model service; Assistants in multi-tenant environments with hard data isolation; LLM observability: tracing, cost accounting, investigating answer-quality incidents; Shipping AI features to external users rather than internal staff only

• Условий в вакансии нет