Загрузи резюме, чтобы видеть мэтчи с вакансиейЗагрузи резюме, чтобы видеть мэтч с вакой
подготовься к отклику
ai-инструменты
Чтобы адаптировать резюме под вакансию или составить сопроводительное письмо, загрузи резюме
описание
A 90-minute working session on a small, real problem over live-streaming data, which we hand you at the start. You work on your own machine, with your own AI tooling, sharing your screen. We are interested in how you frame the problem, what baseline you choose, what you ask us, and how you satisfy yourself that what you built actually works. Fluency with our particular stack is not scored, and there is nothing to prepare. • Tell us about one model or LLM feature you took all the way to production — and about one time your evaluation said it worked and production said otherwise.
Tango is a live-streaming platform with more than 450 million registered users. Its video technology enables people to create live content, engage with fans, and monetize their talent.
задачи
Turn product goals into solutions with the product manager, challenge proposed approaches, suggest alternatives, and determine what to exclude
Build LLM-based features using context engineering, retrieval, tool use, agentic flows, structured outputs, and guardrails; fine-tune or distil models when needed
Train purpose-built models, owning the process from data collection and labelling strategy through training and evaluation
Decide whether to build or buy models based on quality, latency, unit cost, and data-privacy constraints
Ship models to production, including serving under real load, meeting latency and cost budgets, batching, quantization, safe rollout, and integration with backend services and clients
Build and maintain offline evaluation sets, run online A/B tests, monitor drift, regressions, and feedback loops, and roll back models when necessary
Define practices for model evaluation, release gating, prompt and dataset versioning, and build-versus-buy decisions, and share them with the wider engineering organization
Collaborate with the product manager to decide what to build and deliver features from goal to production
требования
Have 5+ years of professional engineering experience, including strong software engineering skills
Have hands-on production experience with LLMs, including retrieval, tool use, agents, structured output, and prompt and context engineering under latency and cost constraints
Have practical experience adapting models through SFT, LoRA, preference tuning, and distillation
Build task-specific evaluation datasets, use LLM-as-judge appropriately, understand its limitations, and assess offline-to-online correlation
Understand LLM failure modes, including hallucination, prompt injection, silent quality drift, and benchmark overfitting; verify model performance rather than relying on assumptions
Have trained and shipped models using production data in classification, ranking, recommendation, computer vision, speech/audio, or abuse and anomaly detection
Own the data path, including collection, labelling strategy and quality, splits, leakage, class imbalance, and their associated risks
Have solid ML fundamentals and be able to debug models
Be fluent in Python and PyTorch and work efficiently with GPUs, understanding training-run costs
Have experience serving models under real load, including throughput, tail latency, batching, caching, quantization, and related trade-offs
Build reproducible pipelines and orchestration, version data, models, and prompts, and identify exactly what was deployed
Monitor production ML for drift and regressions, track feedback loops, and support safe rollback
Be comfortable working with backend, web, Android, and iOS services and clients, with AI tooling and expert support
Use AI coding agents hands-on in daily production work and scope tasks effectively for them
Be comfortable owning features end to end, from design and implementation through evaluation and rollout, and take responsibility for shipped results
Collaborate directly with product managers, designers, and data analysts
Have English proficiency at Intermediate level or higher
Будет плюсом: real-time or streaming inference on live video and audio, on-device or edge models, content moderation, trust and safety or consumer-scale anti-fraud, consumer-feed recommendation and ranking with online experimentation, multimodal video/audio/speech or combined text and image work, serving and training infrastructure such as vLLM, Triton, TensorRT, Ray, or Kubeflow, agentic development workflows such as custom agents, MCP servers, or evals, Java 17+ or Spring Boot experience or readiness to ramp up quickly, live-streaming, video, high-scale consumer products, fintech or communication platforms, Docker, Kubernetes, GCP, BigQuery, or Aerospike
условия
Stock options; every employee is a partner in the company
Competitive salary and performance review bonus
One remote working day per week
Official employment in Cyprus; work permit visa for employees and residence permit for spouses
Medical insurance covering 100% for employees and 75% for family members
Comfortable office in Limassol centre with lunch, dinner, and snacks
Professional growth opportunities, Greek lessons, and sport compensation
Family corporate events
Free attendance at Aris Limassol F.C. matches, with VIP and Sky Box access, events, and merchandise
Gifts for employees
A 90-minute working session on a small, real problem using live-streaming data, provided at the start. Candidates work on their own machine with their own AI tooling and share their screen. Evaluation focuses on problem framing, baseline selection, questions, and verifying that the solution works; familiarity with the company’s stack is not scored, and no preparation is required
