Job Title

Designation - AI EngineerExpereience- 3 to 5 yearsLocation- United Kingdom & IndiaTech Stack: Python, FastAPI, Ray, Delta Lake, Polars, Kubernetes, Hugging Face, Kafka, LLMsAbout UsAt DLT Apps, we’re building a powerful AI-native platform that transforms how financial institutions work with data, documents, and decisions. Across our suite of products — including Terra AI (data migration & doc understanding), Zeta (AI-powered financial advice), and Qkvin (AI-first KYC/AML) — we’re applying LLMs, vision models, and real-time infrastructure to solve real-world financial challenges.You’ll be joining a core engineering team that’s deeply technical, fast-moving, and obsessive about product quality. We build composable, API-first tools that bring ML into the hands of advisors, analysts, and operations teams.What You’ll Be Doing• Designing and building robust APIs to serve AI features like document parsing, entity extraction, summarisation, and classification — all powered by LLMs.• Fine-tuning and integrating open-source LLMs and vision models, optimising for performance, latency, and token efficiency.• Deploying inference workloads on Ray, scaling across GPUs, and integrating them into FastAPI-powered microservices.• Orchestrating workflows using Argo, streaming data from Kafka, and processing it through task runners.• Building data pipelines with Delta Lake, DuckDB, and Polars to feed ML models and power analytics dashboards.• Logging and monitoring model performance in production, setting up observability for drift, quality, and throughput.• Collaborating on internal tools that support rapid ML development — including prompt management, result tracing, and annotation feedback loops.We’d Love to See• Strong Python skills, especially around FastAPI, Pydantic, and ML tooling.• Experience deploying and scaling LLMs in production — bonus if you’ve used vLLM, Transformers, TGI, or NVIDIA Triton.• Solid understanding of Kubernetes, Docker, and cloud-native ML infra (especially for GPU scheduling, job submission, etc.).• Familiarity with vector databases (e.g. Milvus, Qdrant), semantic search, and prompt engineering.• A track record of shipping ML features — not just training models, but putting them into users’ hands.• Bonus: Exposure to financial data, document-heavy workflows, or RegTech.Why You Should Join• Be part of a product-first AI team that’s building for real users — not just papers or benchmarks.• Work across the stack — from inference optimization to data engineering to LLM UX.