Team Lead NLP Engineer

Itcharm; France; Remote; EUR 120,000 - 180,000; Full time

Job description

Imagine a city that never sleeps, and every person in it is an AI you can actually talk to. We are: a chat platform where AI avatars live in cities around the world, from Miami to New York, and build real ongoing relationships with users.

Our primary market is the US, and every avatar is created in-house by our team.

We're an 18+ product, and our business model is usually honest: people pay for premium content when the conversation is worth it. No ads, no tricks, no dark patterns.

Retention, chat depth and revenue all come down to one thing: how good the dialogue is.

That makes ML the core of the company, not a support function. Every improvement you ship (better memory, a stronger persona, fewer loops) shows up in user behavior within days, and you can measure it in A/B tests.

You'll own the ML strategy for our dialogue systems and lead a team of 3 ML engineers. This is a player-coach role: you decide what to build and in what order, tie those bets to business metrics (ARPU, retention, chat depth), and stay hands-on where it matters.

What you'll do

• Own the ML roadmap for dialogue systems and prioritize bets against business metrics.
• Lead 3 ML engineers: set the technical bar, distribute work, hire, give honest performance feedback, and grow your people.
• Drive LLM post-training and model adaptation end to end: SFT, LoRA/QLoRA, preference optimization (DPO/ORPO/SimPO/GRPO), dataset construction.
• Design agent harnesses and tool-using LLM systems: tool calling, structured outputs, routing, retries, memory and context, guardrails.
• Build the evaluation regression suites that layer: offline and pairwise evals, LLM-as-a-judge, correlate with A/B outcomes.
• Reduce dialogue failure modes (loops, contradictions, persona drift, context loss, generic replies) while keeping latency and cost per message under control.
• Stay hands-on (roughly 10–20% of your time): prototypes, debugging agent traces, reviewing your team's work.

What we expect

• Technical degree and a real ML engineering background. You have trained and shipped models yourself, not only managed people who do.
• Experience leading a team of 3–5 ML engineers, including hiring and performance decisions.
• Expert-level Python, a solid understanding of transformer architecture and modern LLM behavior, and hands-on experience with training, fine-tuning and evaluation.
• Production experience with LLM systems: APIs, streaming, batching, fallbacks, cost/latency trade-offs, and observability over traces and transcripts.
• The ability to read fresh research and turn it into a prototype, an eval and a shipped change with measurable impact.
• Fluent Russian; ready to work in CET (±2) hours.

Nice to have

• Distributed training and serving: DeepSpeed / FSDP / Accelerate / Ray; vLLM / TGI / SGLang.
• Agent frameworks and MCP: LangGraph / AutoGen / CrewAI / OpenAI Agents SDK.
• Reward models, LLM-as-a-judge calibration, model routing, mixture of models.
• Quantization (AWQ, GPTQ, bitsandbytes, FP8), KV-cache optimization, LoRA merging, multi-adapter serving.
• Emotion / intent / topic classifiers, conversational recommender systems.
• Experience with role-play, AI companions, character consistency or long-running conversations.
• Coding-agent benchmarks (SWE-bench, Terminal-Bench); voice/realtime and multimodal LLMs.
• SQL/analytics and A/B-testing statistics; enough frontend/backend to ship internal tools.
• Publications in NLP/LLM/agents, open-source contributions.

Who thrives here

Proactive, deeply technical and product-minded people who can read a fresh paper, build a working prototype, train a model, design an eval, debug an agent trace, ship to production, and measure the business impact.

What we offer

• We move fast: ideas become live experiments in days, not quarters.
• Small team, high trust, high ownership - you'll see the direct impact of your work.
• Work from anywhere with our fully remote, full-time setup.
• Standard 28-day annual leave policy.
• 7 yearly wellness days for life admin or recovery - no sick notes needed.
• Referral rewards up to $5000 for helping us hire top talent.
• We cover 50% of costs for training, conferences, and global meetups.
• Subsidized English classes through our corporate discount.
• Health support: receive up to $1,000 annually for medical fees or private insurance if you aren't on the group plan.
• Optimized workspace: we equip you in-office or provide $1000 every three years to perfect your home office or co-working setup.
• Peer-to-peer recognition: earn gratitude bonuses and swap them for swag, massage vouchers, or team adventures.