AI Architect / Tech Lead (mahjong game)
Neurons Lab · Remote · Poland · Portugal · Slovenia · Cyprus · Greece · Serbia · Hungary · Spain · Italy · Slovakia · Lithuania · Albania · Czech Republic · Moldova · Ukraine · Bulgaria · Estonia · Macedonia · Romania ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Part-time (0.5 FTE)
- 3-month contract
- Fully remote
- Python + AWS
- Game AI focus
Join as a hands-on Tech Lead to develop the AI side of a social gaming platform with web3 elements. Your focus is creating a mahjong-playing algorithm that operates at a strong level and explains its moves to users.
You will own the technical architecture, combining a dedicated decision-making model with an LLM reasoning layer. The role requires delivering a solution that respects a two-second response budget per move while supporting various rule sets.
What you'll do
- Design and build the mahjong-playing algorithm using RL, imitation learning, or search methods
- Own the end-to-end technical architecture including the LLM reasoning layer and API contracts
- Optimize inference paths to meet the strict two-second response budget per move
- Define data requirements and build training and calibration pipelines from hand histories
- Create evaluation harnesses to measure play strength and validate explanation quality
- Implement LLM observability using tools like Langfuse for async logging
- Lead sprint work with the AI Engineer and coordinate with the client Product Owner
- Communicate technical decisions and trade-offs directly to the client CTO and engineers
What you'll need
- Six or more years of hands-on ML and AI engineering experience
- Proven work in game AI or sequential decision-making such as RL, MCTS, or self-play
- Expert-level Python skills for building ML systems and robust software engineering
- Experience shipping AI systems in live products with hard latency limits
- Strong background in LLM application engineering including prompt design and guardrails
- Proficiency in low-latency inference techniques like profiling, batching, and caching
- Experience deploying ML workloads on AWS cloud infrastructure
- Background in technology consulting or client-facing delivery roles
Nice to have
- Knowledge of web3 or gaming product contexts
- Experience with imperfect-information games like poker or card games
- Familiarity with AWS Well-Architected frameworks for ML workloads
Stack
- Python
- AWS
- Langfuse
- LLMs
- Reinforcement Learning
How you'll work
- Fully remote position
- Part-time engagement at 0.5 FTE
- Contract duration of three months
- Collaboration in a scrum process with client teams