Machine Learning Lead

About the role

Social Discovery Group is looking for a Machine Learning Lead to own the ML strategy for its dialogue systems and drive the end‑to‑end development of large language model (LLM) solutions. The role combines strategic planning, hands‑on prototyping, and team leadership in a fully remote environment.

Key responsibilities

• Define and prioritize the ML roadmap for dialogue systems, linking projects to business metrics such as ARPU, retention and chat depth.
• Lead a team of three ML engineers: set technical standards, distribute work, hire, manage performance and foster growth.
• Drive LLM post‑training pipelines, including supervised fine‑tuning (SFT), LoRA/QLoRA, and preference optimisation methods (DPO, ORPO, SimPO, GRPO).
• Design tool‑calling agents, structured‑output pipelines, routing, retries, memory management and safety guardrails.
• Build robust evaluation layers with offline, pairwise and LLM‑as‑judge metrics that correlate with A/B test outcomes.
• Identify and eliminate dialogue failure modes (loops, contradictions, persona drift, generic replies) while keeping inference latency and cost low.
• Stay hands‑on (≈10‑20% of time) for prototyping, debugging agent traces and reviewing team deliverables.

Required profile

• Technical degree with a proven ML engineering background – you have trained and shipped models yourself.
• Experience leading a team of 3‑5 ML engineers, including hiring and performance decisions.
• Fluent Russian and able to work within CET (±2) hours.
• Strong ability to translate fresh research into prototypes, evaluations and production‑ready changes with measurable impact.

Required skills

• Expert‑level Python programming.
• Deep understanding of transformer architectures and modern LLM behaviour.
• Hands‑on experience with LLM training, fine‑tuning (SFT, LoRA/QLoRA) and evaluation.
• Production expertise: APIs, streaming, batching, cost/latency trade‑offs, observability of traces and transcripts.
• Familiarity with tools such as vLLM, TGI, SGLang, DeepSpeed, FSDP, Accelerate, quantisation and safety classifiers.

What we offer

• Full‑time remote work opportunity.
• 28 vacation days plus 7 wellness days per year.
• Bonuses up to $5,000 for successful referrals.
• 50% reimbursement for professional training, conferences and meetings.
• Health benefits – up to $1,000 gross per year for insurance or medical expenses.
• Equipment allowance and reimbursement for home‑office or co‑working space costs.
• Internal gamified gratitude system with merchandise, team‑building activities and massage certificates.